<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:wfw="http://wellformedweb.org/CommentAPI/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
	xmlns:slash="http://purl.org/rss/1.0/modules/slash/"
	>

<channel>
	<title>TSql | DBA von Nebenan</title>
	<atom:link href="https://dbavonnebenan.de/category/tsql/feed/" rel="self" type="application/rss+xml" />
	<link>https://dbavonnebenan.de</link>
	<description>Sql Server, Performance &#38; PowerShell Automation</description>
	<lastBuildDate>Fri, 18 Jul 2025 05:26:47 +0000</lastBuildDate>
	<language>de</language>
	<sy:updatePeriod>
	hourly	</sy:updatePeriod>
	<sy:updateFrequency>
	1	</sy:updateFrequency>
	<generator>https://wordpress.org/?v=6.9.4</generator>

<image>
	<url>https://dbavonnebenan.de/wp-content/uploads/2025/04/cropped-www_logo-32x32.png</url>
	<title>TSql | DBA von Nebenan</title>
	<link>https://dbavonnebenan.de</link>
	<width>32</width>
	<height>32</height>
</image> 
	<item>
		<title>Measuring SQL Server Disk Latency Correctly – snapDiskLatency</title>
		<link>https://dbavonnebenan.de/sql-server-disk-latency-en/</link>
					<comments>https://dbavonnebenan.de/sql-server-disk-latency-en/#respond</comments>
		
		<dc:creator><![CDATA[Gabriel]]></dc:creator>
		<pubDate>Thu, 17 Jul 2025 14:21:36 +0000</pubDate>
				<category><![CDATA[Sql Server]]></category>
		<category><![CDATA[TSql]]></category>
		<category><![CDATA[diskio]]></category>
		<category><![CDATA[DISKSPD]]></category>
		<category><![CDATA[eng]]></category>
		<category><![CDATA[latency]]></category>
		<category><![CDATA[PERFORMANCE]]></category>
		<guid isPermaLink="false">https://dbavonnebenan.de/?p=579</guid>

					<description><![CDATA[<p>How to get correct SQL Server Disk Latency values instead of falsified DMV data. Open Source snapDiskLatency solution for long-term analysis.</p>
<p>The post <a href="https://dbavonnebenan.de/sql-server-disk-latency-en/">Measuring SQL Server Disk Latency Correctly – snapDiskLatency</a> first appeared on <a href="https://dbavonnebenan.de">DBA von Nebenan</a>.</p>]]></description>
										<content:encoded><![CDATA[<div style="border: 1px dotted #eb2b14; border-radius: 0%; background-color: #eeeeee; padding-top: var(--wp--preset--spacing--30); padding-left: var(--wp--preset--spacing--30); padding-right: var(--wp--preset--spacing--30); padding-bottom: var(--wp--preset--spacing--30); " class="ub-styled-box ub-bordered-box wp-block-ub-styled-box" id="ub-styled-box-ed9a98aa-0cd4-41e3-8f7e-56d29b672df9">
<p class="has-open-sauce-sans-font-family" style="margin-top:0;margin-right:0;margin-bottom:0;margin-left:0" id="ub-styled-box-bordered-content-">What&#8217;s available here?</p>



<ul style="margin-top:0;margin-right:0;margin-bottom:0;margin-left:0" class="wp-block-list has-open-sauce-sans-font-family">
<li><a href="#DISKSPD" title="">DISKSPD Templates and Wrapper</a></li>



<li><a href="#sqlfilestats" title="">How can I display the latencies of individual database files?</a></li>



<li><a href="#snapdisk" title="">Long-term analysis of latencies with snapDiskLatency</a></li>
</ul>


</div>


<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>


https://github.com/gabrielkoehl/DBAScriptBox


<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)">Intro</h2>



<p>Currently, I&#8217;m involved in a code review at a client. We&#8217;re examining not only the application but also all interfaces. This naturally includes the SQL Server landscape, which I set up at the beginning of the year.</p>



<div class="wp-block-columns is-layout-flex wp-container-core-columns-is-layout-28f84493 wp-block-columns-is-layout-flex">
<div class="wp-block-column is-vertically-aligned-center is-layout-flow wp-block-column-is-layout-flow" style="flex-basis:85%">
<p class="has-open-sauce-sans-font-family">The result was more findings in my direction than expected, and I immediately felt challenged. I was able to refute most of it, some topics were the client&#8217;s responsibility, and one topic caught my interest purely out of curiosity.</p>



<p class="has-text-align-center" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)"><strong>Read- and Write Latencies of TempDB</strong> ( and every other DB )</p>
</div>



<div class="wp-block-column is-vertically-aligned-center is-layout-flow wp-block-column-is-layout-flow" style="flex-basis:15%">
<figure class="wp-block-image aligncenter size-full is-resized"><img fetchpriority="high" decoding="async" width="264" height="329" src="https://dbavonnebenan.de/wp-content/uploads/2025/06/emote-4.png" alt="" class="wp-image-516" style="width:131px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/06/emote-4.png 264w, https://dbavonnebenan.de/wp-content/uploads/2025/06/emote-4-241x300.png 241w" sizes="(max-width: 264px) 100vw, 264px" /></figure>
</div>
</div>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)">Background</h2>



<p>The client doesn&#8217;t operate their own data center, so they&#8217;re dependent on the hoster, who unfortunately isn&#8217;t very communicative. Therefore, I had to get a picture of the totally overcommitted hardware through various other means and challenge the hoster. This was also the reason why I was quite precisely informed about what the hoster&#8217;s hosts and storage systems were capable of delivering.</p>



<p>That&#8217;s why I was very surprised that 32ms Write Latency was listed as a finding. No question, that&#8217;s a bad value for AllFlash NVMe, but it just stood there in the report without differentiation or deeper analysis. Just the statement &#8222;bad storage.&#8220; So I unpacked my tools, took another look at it, and also developed a new solution that I want to share with you.</p>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading" id="DISKSPD" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:0">DISKSPD</h2>



<p style="margin-bottom:var(--wp--preset--spacing--50)"><em><strong>Repo</strong>: perf_diskspeed/DISKSPD</em><br><strong><em>Microsoft</em>:</strong> <a href="https://github.com/microsoft/diskspd" target="_blank" rel="noopener" title="">https://github.com/microsoft/diskspd</a></p>



<p>With DISKSPD, you should always start on unknown systems; often a large portion of problems is hidden here. At this client, for example, a DB with high workload was running on spindles, with latencies in the .2 percentile of 15 seconds latency, plus 14MB throughput. I don&#8217;t need to worry about indexes etc. anymore.</p>



<p>To quickly get results, I built a wrapper for DISKSPD that always executes 3 standard tests:</p>



<ul class="wp-block-list">
<li>SQL Server Workload</li>



<li>Random 64KB Workload</li>



<li>Sequential 64KB Workload</li>
</ul>



<p>The files are stored separately in the _Output subfolder and thus provide quite good hints about storage problems. Now these were, as expected, great values that correspond to NVMe storage. So let&#8217;s continue to SQL Server and see what it tells us. welche NVMe Storage entsprechen. Also weiter zu Sql Server und wir schauen mal was uns dieser so sagt.</p>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading" id="sqlfilestats" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)">sys.dm_io_virtual_file_stats</h2>



<p>SQL Server provides a view where all file metrics are cleanly listed.</p>


<div class="wp-block-syntaxhighlighter-code "><pre class="brush: sql; title: ; notranslate">
select * from sys.dm_io_virtual_file_stats(null,null)
</pre></div>


<ul class="wp-block-list">
<li><a href="https://learn.microsoft.com/en-us/sql/relational-databases/system-dynamic-management-views/sys-dm-io-virtual-file-stats-transact-sql?view=sql-server-ver17">Microsoft Learn</a></li>
</ul>



<figure class="wp-block-image size-full" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)"><img decoding="async" width="1117" height="458" src="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_dmv.png" alt="sys.dm_io_virtual_file_stats" class="wp-image-569" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_dmv.png 1117w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_dmv-300x123.png 300w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_dmv-1024x420.png 1024w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_dmv-768x315.png 768w" sizes="(max-width: 1117px) 100vw, 1117px" /></figure>



<p>Based on this, I built an analysis script that gives me more info and I can also draw further conclusions like ReadWrite Ratio, probable IOPS based on the number of operations, and also quantities. This allows you to build storage systems quite well for specific workloads.</p>



<div class="wp-block-columns is-layout-flex wp-container-core-columns-is-layout-28f84493 wp-block-columns-is-layout-flex">
<div class="wp-block-column is-vertically-aligned-center is-layout-flow wp-block-column-is-layout-flow" style="flex-basis:80%">
<p>And this value also showed me 32ms Write Latency of the TempDB. Then I started wondering, how does this value even arise?</p>
</div>



<div class="wp-block-column is-vertically-aligned-center is-layout-flow wp-block-column-is-layout-flow" style="flex-basis:20%">
<figure class="wp-block-image aligncenter size-full is-resized"><img decoding="async" width="261" height="323" src="https://dbavonnebenan.de/wp-content/uploads/2025/07/emotes_question.png" alt="Question" class="wp-image-575" style="width:125px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/07/emotes_question.png 261w, https://dbavonnebenan.de/wp-content/uploads/2025/07/emotes_question-242x300.png 242w" sizes="(max-width: 261px) 100vw, 261px" /></figure>
</div>
</div>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)">Basic Calculation</h3>



<p>SQL Server measures I/O latency through simple division of cumulative wait times by the number of operations:</p>


<div class="wp-block-syntaxhighlighter-code "><pre class="brush: sql; title: ; notranslate">
-- Read Latency in ms
io_stall_read_ms / num_of_reads = durchschnittliche Read-Latenz

-- Write Latency in ms  
io_stall_write_ms / num_of_writes = durchschnittliche Write-Latenz
</pre></div>


<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<p>That&#8217;s it!!! And now it should be clear that the point-in-time use of this DMV is not really meaningful, on the contrary. The values can, as with the review company that specializes in such processes, lead to false conclusions and action recommendations. Scary.</p>



<p>If the storage system had a hard hiccup since the start of SQL Server, the metrics are contaminated. According to my research, you can&#8217;t flush these, only by taking the DB offline, which I usually avoid in productive environments.</p>



<p>After a few more rounds in my head, the solution came quite quickly. These are all snapshot values, I just need the delta between two values and I have real values. And I can also put the whole thing in a time series and have a long-term analysis. Mindblowing <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f61d.png" alt="😝" class="wp-smiley" style="height: 1em; max-height: 1em;" /></p>



<p>Google also quickly showed corresponding solutions that I used as a base idea. Thanks to Paul Randal ( <a href="https://www.sqlskills.com/blogs/paul/capturing-io-latencies-period-time/" target="_blank" rel="noopener" title="">sqlskills.com</a> )</p>



<p>For analyzing a complete week, this is also not suitable. So I extended the whole thing to a long-term analysis with persistent table.</p>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading" id="snapdisk" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:0">The Solution &#8211;&gt; snapDiskLatency</h2>



<p><em><strong>Repo</strong>: </em>perf_diskspeed/snapDiskLatency</p>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)">What does snapDiskLatency do?</h3>



<p>The problem is well-known &#8211; the standard DMV <code>sys.dm_io_virtual_file_stats</code> shows cumulative values since SQL Server startup. One single storage hiccup weeks ago and the latency values are no longer representative.</p>



<p>This solves the problem with two simple stored procedures:</p>



<p>The first collects the raw data from the DMV hourly and writes it to a permanent table. Why hourly? Shorter intervals lead to unreliable delta calculations &#8211; too little I/O between snapshots makes the values inaccurate. However, this can be individually adjusted since the agent triggers the procedure.</p>



<figure class="wp-block-image size-full" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)"><img loading="lazy" decoding="async" width="1360" height="382" src="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_raw.png" alt="snapDiskLatency collect" class="wp-image-567" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_raw.png 1360w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_raw-300x84.png 300w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_raw-1024x288.png 1024w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_raw-768x216.png 768w" sizes="auto, (max-width: 1360px) 100vw, 1360px" /></figure>



<p>The second does the delta calculations between snapshots and creates the result. Thus, real latency values are generated for defined time periods, without historical hiccups, which can then be compared with other metrics.</p>



<figure class="wp-block-image size-full" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)"><img loading="lazy" decoding="async" width="1298" height="326" src="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_report-2.png" alt="snapDiskLatency report" class="wp-image-571" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_report-2.png 1298w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_report-2-300x75.png 300w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_report-2-1024x257.png 1024w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_report-2-768x193.png 768w" sizes="auto, (max-width: 1298px) 100vw, 1298px" /></figure>



<p>Here we see, for example, in the <a href="https://dbavonnebenan.de/sql-server-truncate-vs-delete-dml-or-ddl-en/" title="TRUNCATE TABLE – DDL vs DML – deep dive in operations and practical tips">TRUNCATE TABLE</a> an outlier (NVMe is also underlying here). On the LAB system there&#8217;s no load, DOXIS is just a table with 2 columns and was artificially worked with workload.<br></p>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)"><strong>The Result</strong></h3>



<p>Instead of possibly contaminated values:</p>



<ul class="wp-block-list">
<li>Average latencies per interval</li>



<li>Read/Write ratios</li>



<li>Breakdown by database and file type</li>
</ul>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)"><strong>Setup</strong></h3>



<p>Create Agent Job, run hourly, done. The complete instructions are in the README, copy &amp; paste &amp; enjoy</p>



<figure class="wp-block-image aligncenter size-full is-resized"><img loading="lazy" decoding="async" width="246" height="470" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_standing_1.png" alt="happy" class="wp-image-330" style="width:125px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_standing_1.png 246w, https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_standing_1-157x300.png 157w" sizes="auto, (max-width: 246px) 100vw, 246px" /></figure>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)">Diagrams</h2>



<p>Since every good proprietary database monitoring tool does this and presents it graphically, I want to at least show one possibility in Excel in the result. For this, the result of the DataSource or Copy must be pivoted and you can easily recognize your daily peaks.</p>



<figure class="wp-block-image size-full" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)"><img loading="lazy" decoding="async" width="1617" height="604" src="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_graph.png" alt="" class="wp-image-572" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_graph.png 1617w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_graph-300x112.png 300w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_graph-1024x382.png 1024w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_graph-768x287.png 768w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_graph-1536x574.png 1536w" sizes="auto, (max-width: 1617px) 100vw, 1617px" /></figure><p>The post <a href="https://dbavonnebenan.de/sql-server-disk-latency-en/">Measuring SQL Server Disk Latency Correctly – snapDiskLatency</a> first appeared on <a href="https://dbavonnebenan.de">DBA von Nebenan</a>.</p>]]></content:encoded>
					
					<wfw:commentRss>https://dbavonnebenan.de/sql-server-disk-latency-en/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>SQL Server Disk Latency richtig messen &#8211; snapDiskLatency</title>
		<link>https://dbavonnebenan.de/sql-server-disk-latency-de/</link>
					<comments>https://dbavonnebenan.de/sql-server-disk-latency-de/#respond</comments>
		
		<dc:creator><![CDATA[Gabriel]]></dc:creator>
		<pubDate>Thu, 17 Jul 2025 14:06:35 +0000</pubDate>
				<category><![CDATA[Sql Server]]></category>
		<category><![CDATA[TSql]]></category>
		<category><![CDATA[DE]]></category>
		<category><![CDATA[diskio]]></category>
		<category><![CDATA[DISKSPD]]></category>
		<category><![CDATA[latency]]></category>
		<category><![CDATA[PERFORMANCE]]></category>
		<guid isPermaLink="false">https://dbavonnebenan.de/?p=563</guid>

					<description><![CDATA[<p>Wie du richtige SQL Server Disk Latency Werte erhältst statt verfälschter DMV-Daten. Open Source snapDiskLatency Lösung für Langzeitanalyse.</p>
<p>The post <a href="https://dbavonnebenan.de/sql-server-disk-latency-de/">SQL Server Disk Latency richtig messen – snapDiskLatency</a> first appeared on <a href="https://dbavonnebenan.de">DBA von Nebenan</a>.</p>]]></description>
										<content:encoded><![CDATA[<div style="border: 1px dotted #eb2b14; border-radius: 0%; background-color: #eeeeee; padding-top: var(--wp--preset--spacing--30); padding-left: var(--wp--preset--spacing--30); padding-right: var(--wp--preset--spacing--30); padding-bottom: var(--wp--preset--spacing--30); " class="ub-styled-box ub-bordered-box wp-block-ub-styled-box" id="ub-styled-box-ba0a6049-a2c1-44fc-b531-47f63f78c8b4">
<p style="margin-top:0;margin-right:0;margin-bottom:0;margin-left:0" id="ub-styled-box-bordered-content-">Was erwartet dich hier?</p>



<ul style="margin-top:0;margin-right:0;margin-bottom:0;margin-left:0" class="wp-block-list">
<li><a href="#DISKSPD" title="DISKSPD Templates und Wrapper">DISKSPD Templates und Wrapper</a></li>



<li><a href="#sqlfilestats" title="">Wie kann ich die Latenzen der einzelnen Datenbank Files anzeigen lassen?</a></li>



<li><a href="#snapdisk" title="Langzeit Analyse der Latenzen!!!">Langzeit Analyse der Latenzen!!!</a></li>
</ul>


</div>


<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>


https://github.com/gabrielkoehl/DBAScriptBox


<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)">Intro</h2>



<p>Aktuell bin ich bei einem Kunden an einem Code Review beteiligt. Dabei beleuchten wir neben der Applikation auch alle Schnittstellen.. Dazu gehört natürlich auch die Sql Server Landschaft, welche ich Anfang des Jahres aufgesetzt habe.</p>



<div class="wp-block-columns is-layout-flex wp-container-core-columns-is-layout-28f84493 wp-block-columns-is-layout-flex">
<div class="wp-block-column is-vertically-aligned-center is-layout-flow wp-block-column-is-layout-flow" style="flex-basis:85%">
<p>Das Ergebnis waren doch mehr Findings in meine Richtung als erwartet und ich fühlte mich sofort gechallenged. Das Meiste konnte ich widerlegen, bei einigen Themen war der Kunde in der Pflicht und ein Thema hat mich rein interessenhalber gepackt.</p>



<p class="has-text-align-center" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)"><strong>Read- und Write Latenzen der TempDB</strong> ( und jeder anderen DB )</p>
</div>



<div class="wp-block-column is-vertically-aligned-center is-layout-flow wp-block-column-is-layout-flow" style="flex-basis:15%">
<figure class="wp-block-image aligncenter size-full is-resized"><img loading="lazy" decoding="async" width="264" height="329" src="https://dbavonnebenan.de/wp-content/uploads/2025/06/emote-4.png" alt="" class="wp-image-516" style="width:131px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/06/emote-4.png 264w, https://dbavonnebenan.de/wp-content/uploads/2025/06/emote-4-241x300.png 241w" sizes="auto, (max-width: 264px) 100vw, 264px" /></figure>
</div>
</div>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)">Vorgeschichte</h2>



<p>Der Kunde betreibt kein eigenes Rechenzentrum, ist somit vom Hoster abhängig, welcher leider nicht sehr kommunikativ ist. Somit musste ich mir auf diverse andere Wege ein Bild von der total overcommitteten Hardware machen und den Hoster Challengen. Dies war auch der Grund, warum ich ziemlich genau im Bilde war, was die Hosts und Storage Systeme des Hosters im Stande waren zu leisten.</p>



<p>Deshalb hat es mich doch sehr gewundert, dass als Finding 32ms Write Latency aufgeführt waren. Keine Frage, das ist ein schlechter Wert für AllFlash NVMe aber der stand da einfach so im Report, ohne Differenzierung oder tiefere Analysen. Nur die Aussage schlechter Storage. Also habe ich meine Werkzeuge ausgepackt, mir das noch mal angeschaut und auch eine neue Lösung entwickelt, die ich mit euch teilen möchte.</p>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading" id="DISKSPD" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:0">DISKSPD</h2>



<p style="margin-bottom:var(--wp--preset--spacing--50)"><em><strong>Repo</strong>: perf_diskspeed/DISKSPD</em><br><strong><em>Microsoft</em>:</strong> <a href="https://github.com/microsoft/diskspd">https://github.com/microsoft/diskspd</a></p>



<p>Mit DISKSPD sollte man auf unbekannten Systemen immer anfangen, oft liegt hier schon ein großer Anteil an Problemen versteckt. Bei diesem Kudnen lief zum Beispiel eine DB mit hohem Workload auf Spindeln, mit Latenzen in der .2 Perzentile mit 15 Sekunden Latenz, dazu 14MB Durchsatz. Da brauche ich mir um Indexe usw. keine Gedanken mehr machen.</p>



<p>Um schnell an Ergebnisse zu kommen, habe ich mir einen Wrapper für DISKSP gebaut, welcher immer 3 Standardtests ausführt</p>



<ul class="wp-block-list">
<li>Sql Server Workload</li>



<li>Random 64KB Workload</li>



<li>Sequenziell 64KB Workload</li>
</ul>



<p>Die Dateien werden separat im Unterordner _Output gespeichert und ermöglichen somit recht gute Hinweise auf Storage-Probleme. Nun waren das wie erwartet super Werte, welche NVMe Storage entsprechen. Also weiter zu Sql Server und wir schauen mal was uns dieser so sagt.</p>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading" id="sqlfilestats" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)">sys.dm_io_virtual_file_stats</h2>



<p>Sql Server bietet eine View, in der alle Metriken der Files sauber aufgelistet werden.</p>


<div class="wp-block-syntaxhighlighter-code "><pre class="brush: sql; title: ; notranslate">
select * from sys.dm_io_virtual_file_stats(null,null)
</pre></div>


<ul class="wp-block-list">
<li><a href="https://learn.microsoft.com/en-us/sql/relational-databases/system-dynamic-management-views/sys-dm-io-virtual-file-stats-transact-sql?view=sql-server-ver17" target="_blank" rel="noopener" title="">microsoft.com</a></li>
</ul>



<figure class="wp-block-image size-full" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)"><img loading="lazy" decoding="async" width="1117" height="458" src="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_dmv.png" alt="sys.dm_io_virtual_file_stats" class="wp-image-569" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_dmv.png 1117w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_dmv-300x123.png 300w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_dmv-1024x420.png 1024w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_dmv-768x315.png 768w" sizes="auto, (max-width: 1117px) 100vw, 1117px" /></figure>



<p>Darauf basierend habe ich ein Analyse Script gebaut, welches mir mehr Infos gibt und ich auch weitere Rückschlüsse wie ReadWrite Ratio, wahrscheinliche IOPS auf Basis der Anzahl der Vorgänge und auch Mengen. Damit kann man schon ganz gut Storage Systeme zu speziellen Workloads aufbauen.</p>



<div class="wp-block-columns is-layout-flex wp-container-core-columns-is-layout-28f84493 wp-block-columns-is-layout-flex">
<div class="wp-block-column is-vertically-aligned-center is-layout-flow wp-block-column-is-layout-flow" style="flex-basis:80%">
<p>Und auch dieser Wert hat mir 32ms Write Latency der TempDB angezeigt. Dann habe ich angefangen mich zu fragen, wie entstehet dieser Wert überhaupt? </p>
</div>



<div class="wp-block-column is-vertically-aligned-center is-layout-flow wp-block-column-is-layout-flow" style="flex-basis:20%">
<figure class="wp-block-image aligncenter size-full is-resized"><img loading="lazy" decoding="async" width="261" height="323" src="https://dbavonnebenan.de/wp-content/uploads/2025/07/emotes_question.png" alt="Question" class="wp-image-575" style="width:125px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/07/emotes_question.png 261w, https://dbavonnebenan.de/wp-content/uploads/2025/07/emotes_question-242x300.png 242w" sizes="auto, (max-width: 261px) 100vw, 261px" /></figure>
</div>
</div>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)">Grundlegende Berechnung der Werte</h3>



<p>SQL Server misst I/O-Latenz durch einfache Division der kumulativen Wartezeiten durch die Anzahl der Operationen:</p>


<div class="wp-block-syntaxhighlighter-code "><pre class="brush: sql; title: ; notranslate">
-- Read Latency in ms
io_stall_read_ms / num_of_reads = durchschnittliche Read-Latenz

-- Write Latency in ms  
io_stall_write_ms / num_of_writes = durchschnittliche Write-Latenz
</pre></div>


<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<p>Thats it!!! Und nun sollte klar sein, dass die punktuelle Nutzung dieser DMV nicht wirklich aussagekräftig ist, im Gegenteil. Die Werte können wie bei dem Review Unternehmen, was sich auf solche Prozesse spezialisiert hat, zu falschen Rückschlüssen und Handlungsempfehlungen führen. Gruselig.</p>



<p>Wenn das Storagesystem seit dem Start von Sql Server mal hart Schluckauf hatte, sind die Metriken verunreiningt. Laut meinen Recherchen kann man diese nicht flushen, nur durch das Offline nehmen der DB, was ich üblicherweise in produktiven Umgebungen vermeide.</p>



<p>Nach ein paar weiteren Runden in meinem Kopf kam recht schnell die Lösung. Das sind alles SnapShot Values, ich brauche ja nur das Delta zwischen zwei Werten und schon habe ich reale Werte. Und das Ganze kann ich auch noch in eine Zeitreihe packen und habe eine Langzeitanalyse.  Mindblowing <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f61d.png" alt="😝" class="wp-smiley" style="height: 1em; max-height: 1em;" /></p>



<p>Google hat auch schnell entsprechende Lösungen aufgezeigt, die ich als Basis Idee genutzt habe. Danke an Paul Randal ( <a href="https://www.sqlskills.com/blogs/paul/capturing-io-latencies-period-time/">sqlskills.com</a> )</p>



<p>Für die Analyse einer kompletten Woche ist dies aber auch nicht geeignet. Somit habe ich das Ganze auf eine Langzeitanalyse mit persistenten Table erweitert.</p>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading" id="snapdisk" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:0">Die Lösung &#8211;&gt; snapDiskLatency</h2>



<p><em><strong>Repo</strong>: </em>perf_diskspeed/snapDiskLatency</p>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)">Was macht snapDiskLatency?</h3>



<p>Das Problem ist bekannt &#8211; die Standard-DMV <code>sys.dm_io_virtual_file_stats</code> zeigt kumulative Werte seit dem SQL Server Start. Ein einziger Storage-Schluckauf vor Wochen und die Latenzwerte sind nicht mehr repräsentativ. </p>



<p><strong>snapDiskLatency</strong> löst das Problem mit zwei simplen Stored Procedures:</p>



<p>Die Erste sammelt stündlich die Rohdaten aus der DMV und schreibt sie in einen permanenten Table. Warum stündlich? Kürzere Intervalle führen zu unzuverlässigen Delta-Berechnungen &#8211; zu wenig I/O zwischen den Snapshots macht die Werte ungenau. Dies kann aber individuell angepasst werden, da der Agent die Prozedur triggert.</p>



<figure class="wp-block-image size-full" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)"><img loading="lazy" decoding="async" width="1360" height="382" src="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_raw.png" alt="snapDiskLatency collect" class="wp-image-567" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_raw.png 1360w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_raw-300x84.png 300w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_raw-1024x288.png 1024w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_raw-768x216.png 768w" sizes="auto, (max-width: 1360px) 100vw, 1360px" /></figure>



<p>Die Zweite macht die Delta-Berechnungen zwischen den Snapshots und erstellt das Resultset. Somit werden reale Latenzwerte für definierte Zeiträume erzeugt, ohne historischen Schluckauf, welche dann mit anderen Metriken abgeglichen werden können.</p>



<figure class="wp-block-image size-full" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)"><img loading="lazy" decoding="async" width="1298" height="326" src="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_report-2.png" alt="snapDiskLatency report" class="wp-image-571" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_report-2.png 1298w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_report-2-300x75.png 300w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_report-2-1024x257.png 1024w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_report-2-768x193.png 768w" sizes="auto, (max-width: 1298px) 100vw, 1298px" /></figure>



<p>Hier sehen wir zum Beispiel im <a href="https://dbavonnebenan.de/sql-server-truncate-vs-delete-dml-or-ddl-de/" title="TRUNCATE TABLE Operationen im Detail – ROLLBACK ist doch möglich">TRUNCATE Table</a> einen Aureißer ( hier liegt auch NVMe drunter ). Auf dem LAB System ist keine Last, DOXIS ist nur ein Table mit 2 Spalten und wurde künstlich mit Workload bearbeitet.</p>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)"><strong>Das Ergebnis</strong></h3>



<p>Statt möglicherweise verunreinigten Werten</p>



<ul class="wp-block-list">
<li>Durchschnittliche Latenzen pro Intervall</li>



<li>Read/Write-Verhältnisse</li>



<li>Aufschlüsselung nach Datenbank und File-Type</li>
</ul>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)"><strong>Setup</strong></h3>



<p>Agent Job erstellen, stündlich laufen lassen, fertig. Die komplette Anleitung steht in der README, copy &amp; paste &amp; freuen</p>



<figure class="wp-block-image aligncenter size-full is-resized"><img loading="lazy" decoding="async" width="246" height="470" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_standing_1.png" alt="happy" class="wp-image-330" style="width:125px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_standing_1.png 246w, https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_standing_1-157x300.png 157w" sizes="auto, (max-width: 246px) 100vw, 246px" /></figure>



<div style="height:20px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)">Diagramme</h2>



<p>Da dies auch jedes gute propritäre DatabaseMonitoringTool macht und hübsch grafisch darstellt, will ich zumindest eine Möglichkeit in Excel im Ergebnis zeigen. Dazu muss das Ergebnis der DataSource oder Copy pivotiert werden und schon kann man gut seine Daily Peaks erkennen.</p>



<figure class="wp-block-image size-full" style="margin-top:var(--wp--preset--spacing--50);margin-bottom:var(--wp--preset--spacing--50)"><img loading="lazy" decoding="async" width="1617" height="604" src="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_graph.png" alt="" class="wp-image-572" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_graph.png 1617w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_graph-300x112.png 300w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_graph-1024x382.png 1024w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_graph-768x287.png 768w, https://dbavonnebenan.de/wp-content/uploads/2025/07/disklat_graph-1536x574.png 1536w" sizes="auto, (max-width: 1617px) 100vw, 1617px" /></figure><p>The post <a href="https://dbavonnebenan.de/sql-server-disk-latency-de/">SQL Server Disk Latency richtig messen – snapDiskLatency</a> first appeared on <a href="https://dbavonnebenan.de">DBA von Nebenan</a>.</p>]]></content:encoded>
					
					<wfw:commentRss>https://dbavonnebenan.de/sql-server-disk-latency-de/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>TRUNCATE TABLE &#8211; DDL vs DML &#8211; deep dive in operations and practical tips</title>
		<link>https://dbavonnebenan.de/sql-server-truncate-vs-delete-dml-or-ddl-en/</link>
					<comments>https://dbavonnebenan.de/sql-server-truncate-vs-delete-dml-or-ddl-en/#respond</comments>
		
		<dc:creator><![CDATA[Gabriel]]></dc:creator>
		<pubDate>Mon, 26 May 2025 13:14:32 +0000</pubDate>
				<category><![CDATA[Sql Server]]></category>
		<category><![CDATA[TSql]]></category>
		<category><![CDATA[eng]]></category>
		<category><![CDATA[PERFORMANCE]]></category>
		<category><![CDATA[ROLLBACK]]></category>
		<category><![CDATA[TABLE]]></category>
		<category><![CDATA[TRANSACTION]]></category>
		<category><![CDATA[TRUNCATE]]></category>
		<guid isPermaLink="false">https://dbavonnebenan.de/?p=323</guid>

					<description><![CDATA[<p>Recently, I stumbled across a LinkedIn post about truncate vs. delete, just like many others out there. Typically, random members post simple SQL Server tips, nicely packaged in graphics and kept general. Although for beginners this is absolutely legitimate &#8211; since these simple entry points can help many newcomers understand and provide a great starting [&#8230;]</p>
<p>The post <a href="https://dbavonnebenan.de/sql-server-truncate-vs-delete-dml-or-ddl-en/">TRUNCATE TABLE – DDL vs DML – deep dive in operations and practical tips</a> first appeared on <a href="https://dbavonnebenan.de">DBA von Nebenan</a>.</p>]]></description>
										<content:encoded><![CDATA[<p>Recently, I stumbled across a LinkedIn post about truncate vs. delete, just like many others out there. Typically, random members post simple SQL Server tips, nicely packaged in graphics and kept general. Although for beginners this is absolutely legitimate &#8211; since<strong> </strong>these simple entry points can help many newcomers understand and provide a great starting point &#8211; nevertheless, the fact that the same topics then go viral across various accounts is clearly another matter entirely.</p>



<p>Anyway. If you&#8217;re a bit further along in the topic and already know that the pronunciation isn&#8217;t &#8222;SQL&#8220; but &#8222;sequel&#8220; <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f60a.png" alt="😊" class="wp-smiley" style="height: 1em; max-height: 1em;" />, you should fundamentally question these posts, as they mostly generalize.</p>



<p>This was also the case in a post about the top 3 SQL mistakes that &#8222;everyone&#8220; makes. And here we are, getting to the point.</p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<figure class="wp-block-image aligncenter size-full"><a href="https://www.linkedin.com/posts/samiul-fahim_sql-dataengineering-techtips-activity-7328642722653786112-OBfj?utm_source=share&amp;utm_medium=member_desktop&amp;rcm=ACoAACt1yIMBIZu3k0gdHLx7c_luIAnHYYBMtuc"><img loading="lazy" decoding="async" width="391" height="158" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-26_09-15.png" alt="TRUNCATE cant rolled back" class="wp-image-324" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-26_09-15.png 391w, https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-26_09-15-300x121.png 300w" sizes="auto, (max-width: 391px) 100vw, 391px" /></a></figure>



<p class="has-text-align-center"><a href="https://www.linkedin.com/posts/samiul-fahim_sql-dataengineering-techtips-activity-7328642722653786112-OBfj?utm_source=share&amp;utm_medium=member_desktop&amp;rcm=ACoAACt1yIMBIZu3k0gdHLx7c_luIAnHYYBMtuc" target="_blank" rel="noreferrer noopener">LinkedIn</a></p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<p>The same happened with <a href="https://www.linkedin.com/in/vedrankesegic/">Vedran Kesegic</a>, who gave the hint that TRUNCATE is apparently rollback-capable when encapsulated in a transaction. Then <a href="https://www.linkedin.com/in/saastamoinen/">Carsten Saastamoinen-Jakobsen</a> jumped in and things got wild. So wild that the whole topic hooked me so much that I did a deep dive into it. Furthermore, so much knowledge dropped in this exchange that it would be a shame if it disappeared in a comment section.</p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">What to expect here</h2>



<p>I compared the exchange and looked at the core statements. I also texted separately with both of them a bit more, and Carsten gave me excerpts from one of his publications where the topic is covered technically.</p>



<p>Technically speaking, both agree, even if they don&#8217;t say it <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f923.png" alt="🤣" class="wp-smiley" style="height: 1em; max-height: 1em;" />. But what was really wild in the end was the question:</p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<figure class="wp-block-image aligncenter size-full"><img loading="lazy" decoding="async" width="595" height="159" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/ddl_dml.png" alt="DML or DDL ??" class="wp-image-326" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/ddl_dml.png 595w, https://dbavonnebenan.de/wp-content/uploads/2025/05/ddl_dml-300x80.png 300w" sizes="auto, (max-width: 595px) 100vw, 595px" /></figure>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<div class="wp-block-columns is-layout-flex wp-container-core-columns-is-layout-28f84493 wp-block-columns-is-layout-flex">
<div class="wp-block-column is-layout-flow wp-block-column-is-layout-flow" style="flex-basis:100%">
<div class="wp-block-group is-nowrap is-layout-flex wp-container-core-group-is-layout-6c531013 wp-block-group-is-layout-flex">
<figure class="wp-block-image size-full"><img loading="lazy" decoding="async" width="246" height="470" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_standing_1.png" alt="" class="wp-image-330" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_standing_1.png 246w, https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_standing_1-157x300.png 157w" sizes="auto, (max-width: 246px) 100vw, 246px" /></figure>



<p>I can say it upfront. This question is an philosophical one. I was able to form an opinion based on my research, more on that later, but basically it&#8217;s up to everyone to decide for themselves.</p>
</div>
</div>
</div>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">The Discussion &#8211; DML or DDL ?</h2>



<h3 class="wp-block-heading">Vedran Kesegic</h3>



<p><strong>Core Position: TRUNCATE is DDL</strong></p>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p>&#8222;TRUNCATE modifies internal system tables sys.sysallocunits and sys.sysrowsets. DELETE modifies indexes and stats&#8220;</p>
</blockquote>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p>&#8222;DML writes data (rows in user tables), while DDL writes metadata (rows in system tables, data that describes other/user objects). Truncate happens to write metadata (system tables) and thus is classified as DDL.&#8220;</p>
</blockquote>



<p><strong>Technical Arguments</strong></p>



<ul class="wp-block-list">
<li>Uses SCH-M (Schema Modification) lock</li>



<li>Modifies system metadata tables</li>



<li>Microsoft documentation classifies it as DDL</li>



<li>Logs page deallocations, not row deletions</li>
</ul>



<p><strong>Key Statement</strong></p>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p>&#8222;TRUNCATE modifies internal system tables sys.sysallocunits and sys.sysrowsets. DELETE modifies indexes and stats &#8211; see attached&#8220;</p>
</blockquote>



<div style="height:50px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">Carsten Saastamoinen-Jakobsen</h3>



<p><strong>Core Position: TRUNCATE is DML</strong></p>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p>&#8222;It is not an argument to call TRUNCATE a DDL statement because a certain lock is used! When statements are categorized, it must be based on what happens and not how a given database system solves the task.&#8220;</p>
</blockquote>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p>&#8222;We may have to recognize that a previous error categorization should not continue to call TRUNCATE a DDL, but recognize that it is a DML.&#8220;</p>
</blockquote>



<p><strong>Logical Arguments</strong></p>



<ul class="wp-block-list">
<li>Data is deleted, structure is preserved</li>



<li>No table definition changes occur</li>



<li>Follows semantic rules of data manipulation</li>



<li>Can delete only some partitions (not all data)</li>
</ul>



<p><strong>Key Statement</strong></p>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p>&#8222;I say that MS is right in the documentations because they are not telling that TRUNCATE is as DDL, because they know that TRUNCATE is a DML!&#8220;</p>



<p><mark style="background-color:rgba(0, 0, 0, 0)" class="has-inline-color has-denary-color"><em>NOTE Gabriel <br></em></mark><em>I think this is really a bit questionable because the documentation is absolutely clear</em> <em>on this point</em> &#8211;&gt; <strong><em>DDL </em></strong>(<a href="https://learn.microsoft.com/en-us/sql/t-sql/statements/statements?view=sql-server-ver17#data-definition-language" target="_blank" rel="noreferrer noopener">microsoft.com</a>)<br></p>
</blockquote>



<p><strong>Quote from publication</strong></p>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p>&#8222;As of version 2016, it is possible to delete only some of the data in a table with TRUNCATE, so it is even more cor­rect to call it manipulation than definition, because it is changing some of the data but not changing the structure.&#8220;</p>
</blockquote>



<div style="height:50px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">Points of Agreement</h3>



<ul class="wp-block-list">
<li>TRUNCATE can be rolled back in a transaction
<ul class="wp-block-list">
<li>general demo at <a href="https://blog.sqlauthority.com/2010/03/04/sql-server-rollback-truncate-command-in-transaction/" target="_blank" rel="noreferrer noopener">sqlauthority.com</a></li>



<li>deep dive demo <a href="https://www.sqlshack.com/the-internals-of-sql-truncate-and-sql-delete-statements/" target="_blank" rel="noreferrer noopener">sqlshack.com</a></li>
</ul>
</li>



<li>TRUNCATE is a logged operation</li>



<li>TRUNCATE is faster than DELETE for large datasets</li>



<li>Both understand the technical implementation</li>
</ul>



<div style="height:50px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">Core Disagreement</h3>



<ul class="wp-block-list">
<li><strong>Classification criteria</strong>
<ul class="wp-block-list">
<li>Implementation details</li>
</ul>
</li>



<li><strong>Authority</strong>
<ul class="wp-block-list">
<li>Vendor documentation (Vedran) vs. Logical reasoning (Carsten)</li>
</ul>
</li>



<li><strong>Perspective</strong>
<ul class="wp-block-list">
<li>How it works internally (Vedran) vs. What it does functionally (Carsten)</li>
</ul>
</li>
</ul>



<div style="height:50px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">My 50 Cent</h3>



<p>So as you can see, technically it fits, but how do you view the whole thing? Maybe it&#8217;s good that I&#8217;m not preloaded with as much knowledge as both of them because from my perspective it&#8217;s quite clear: <strong>DDL!!</strong></p>



<div class="wp-block-group is-nowrap is-layout-flex wp-container-core-group-is-layout-6c531013 wp-block-group-is-layout-flex">
<figure class="wp-block-image size-full is-resized"><img loading="lazy" decoding="async" width="292" height="342" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_1.png" alt="" class="wp-image-344" style="width:98px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_1.png 292w, https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_1-256x300.png 256w" sizes="auto, (max-width: 292px) 100vw, 292px" /></figure>



<p>Contrary to Carsten&#8217;s statement that no system tables are changed, Vedran&#8217;s statement is absolutely correct. See why&#8230; </p>
</div>



<p>&#8230; and of course because MS says it&#8217;s DDL <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f60a.png" alt="😊" class="wp-smiley" style="height: 1em; max-height: 1em;" /> ( <a href="https://learn.microsoft.com/en-us/sql/t-sql/statements/statements?view=sql-server-ver17#data-definition-language" target="_blank" rel="noreferrer noopener">microsoft.com</a> )</p>



<p>And partitioning is DML… I don&#8217;t know, here too it&#8217;s at least system columns that define the partitioning.</p>



<p></p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">How TRUNCATE works</h2>



<p>I&#8217;d like to say I&#8217;ll describe it simply, but that&#8217;s just my normal description. The topic is really complex <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/270c.png" alt="✌" class="wp-smiley" style="height: 1em; max-height: 1em;" /></p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">1 &#8211; TRUNCATE Command Initiated</h3>


<div class="wp-block-syntaxhighlighter-code "><pre class="brush: sql; title: ; notranslate">
TRUNCATE TABLE dbo.MyTable;
</pre></div>


<ul class="wp-block-list">
<li><strong>First action</strong>: Acquire <strong>SCH-M (Schema Modification)</strong> lock on entire table</li>



<li><strong>Effect</strong>: Table is completely locked &#8211; no reads, no writes, nothing</li>
</ul>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">2 &#8211; Identify Extents</h3>



<ul class="wp-block-list">
<li>DEEP DIVE Extents &#8211;&gt; <a href="https://learn.microsoft.com/en-us/sql/relational-databases/pages-and-extents-architecture-guide?view=sql-server-ver17" target="_blank" rel="noreferrer noopener">microsoft.com</a></li>



<li>SQL Server reads IAM pages</li>



<li>Lists all extents belonging to the table</li>



<li><strong>Still holding</strong>: SCH-M lock on table</li>
</ul>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">3-  Lock Individual Extents</h3>



<ul class="wp-block-list">
<li>Places <strong>X (Exclusive)</strong> locks on each extent</li>



<li>These are regular data locks, NOT schema locks</li>



<li><strong>Now holding</strong>:
<ul class="wp-block-list">
<li>SCH-M lock on table (prevents access)</li>



<li>X locks on extents (marks them for deallocation)</li>
</ul>
</li>
</ul>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">4 &#8211; Log the Operation</h3>



<ul class="wp-block-list">
<li>Writes to transaction log: &#8222;Deallocate extents 8:8, 8:16, etc.&#8220;</li>



<li><strong>Still holding</strong>: Both lock types</li>
</ul>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">5a &#8211; If COMMIT</h3>



<ul class="wp-block-list">
<li>Updates IAM &#8211; marks extents as free</li>



<li><strong>Releases</strong> all X locks on extents</li>



<li><strong>Releases</strong> SCH-M lock on table</li>



<li>Table is accessible again</li>
</ul>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">5b &#8211; If ROLLBACK</h3>



<ul class="wp-block-list">
<li>Discards deallocation markers</li>



<li><strong>Releases</strong> all X locks on extents</li>



<li><strong>Releases</strong> SCH-M lock on table</li>



<li>Everything unchanged</li>
</ul>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">Why Two Different Lock Types?</h3>



<p><strong>SCH-M Lock:</strong></p>



<ul class="wp-block-list">
<li>&#8222;Nobody touch this table AT ALL&#8220;</li>



<li>Typical for DDL operations</li>



<li>Prevents even metadata queries</li>
</ul>



<p><strong>X Locks on Extents:</strong></p>



<ul class="wp-block-list">
<li>&#8222;These specific data blocks are being modified&#8220;</li>



<li>Typical for data operations</li>



<li>Ensures extents aren&#8217;t reused until commit/rollback by other sessions</li>
</ul>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">My humble Opinion</h3>



<p>Given this knowledge, the case is actually clear for me, but how do you see it? Indeed, here no user data is being manipulated; instead, system-level assignments are being adjusted. Consequently, I&#8217;ll allow myself, despite my new half-knowledge, to make the following comparison:</p>



<div class="wp-block-group is-nowrap is-layout-flex wp-container-core-group-is-layout-6c531013 wp-block-group-is-layout-flex">
<figure class="wp-block-image size-full is-resized"><img loading="lazy" decoding="async" width="307" height="328" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_2.png" alt="" class="wp-image-345" style="width:131px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_2.png 307w, https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_2-281x300.png 281w" sizes="auto, (max-width: 307px) 100vw, 307px" /></figure>



<p>&#8222;I also don&#8217;t say I&#8217;m deleting text in a Word document when I quick format the disk and only delete the block allocation.&#8220;</p>
</div>



<p><br>Yes, I think that&#8217;s a nice comparison&#8230; Case closed, now a few technically interesting aspects and solutions</p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">Technical Part, Hacks and How2</h2>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<figure class="wp-block-image aligncenter size-full is-resized"><img loading="lazy" decoding="async" width="643" height="388" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/meme_trunc.jpg" alt="" class="wp-image-348" style="width:325px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/meme_trunc.jpg 643w, https://dbavonnebenan.de/wp-content/uploads/2025/05/meme_trunc-300x181.jpg 300w" sizes="auto, (max-width: 643px) 100vw, 643px" /></figure>



<p class="has-text-align-center has-medium-font-size">&#8230; practical, not technical <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f609.png" alt="😉" class="wp-smiley" style="height: 1em; max-height: 1em;" /></p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">TRUNCATE with Survivor Table (Temporary Table)</h3>



<p>Data that is still required is simply moved to a temporary table and retrieved after TRUNCATE</p>


<div class="wp-block-syntaxhighlighter-code "><pre class="brush: sql; title: ; notranslate">
-- Step 1: Create temp table with rows to keep
	SELECT * INTO #SurvivorRows 
	FROM 
		MyTable 
	WHERE
		DateColumn &gt;= &#039;2024-01-01&#039;

-- Step 2: Truncate the entire table
	TRUNCATE TABLE MyTable

-- Step 3: Re-insert survivor rows
	INSERT INTO MyTable
	SELECT * FROM #SurvivorRows

-- Step 4: Clean up
	DROP TABLE #SurvivorRows
</pre></div>


<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">TRUNCATE with PARTITION option (2016++)</h3>



<p>Since SQL Server 2016 (every edition) it is possible to specify partitions or ranges of partitions</p>


<div class="wp-block-syntaxhighlighter-code "><pre class="brush: sql; title: ; notranslate">
-- Truncate specific partitions
	TRUNCATE TABLE MyPartitionedTable 
	     WITH (PARTITIONS (2, 4 TO 6))

-- Check partition status
	SELECT 
		 partition_number
		,rows 
	FROM
		sys.partitions 
	WHERE 
		OBJECT_NAME(OBJECT_ID) = &#039;MyPartitionedTable&#039;
</pre></div>


<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">TRUNCATE is always a transaction&#8230;</h3>



<p>&#8230;. and must be considered during RESTORING. Carsten illustrated this whole thing very nicely in his publication. I&#8217;m just describing it, that has to suffice at this point <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f61d.png" alt="😝" class="wp-smiley" style="height: 1em; max-height: 1em;" /></p>



<p>Even though we only write <strong>TRUNCATE TABLE</strong>, SQL Server still uses <strong>BEGIN</strong> and <strong>COMMIT TRANSACTION</strong>, which of course is recorded in the TLog. Accordingly, during a POINT IN TIME RECOVERY, the data in the table remains available until the TLog with the corresponding TRANSACTION is restored, or the time of the TRUNCATE transaction has been exceeded.</p>



<p>So you can also restore the data <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f44c.png" alt="👌" class="wp-smiley" style="height: 1em; max-height: 1em;" /></p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">Continuously read Transaction log file data with fn_dblog and fn_dump_dblog</h3>



<p>Vedran wrote in one of his posts that there&#8217;s an undocumented function where you can read log records at runtime. There&#8217;s also a very cool article about this on sqlshack.com</p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<p class="has-text-align-center"><em><mark style="background-color:rgba(0, 0, 0, 0)" class="has-inline-color has-septenary-color">NOT recommended for production but only for test / education.</mark></em></p>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>


https://www.sqlshack.com/how-to-continuously-read-transaction-log-file-data-directly-in-a-sql-server-database-with-fn_dblog-and-fn_dump_dblog/


<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">Finally</h2>



<div class="wp-block-group is-nowrap is-layout-flex wp-container-core-group-is-layout-6c531013 wp-block-group-is-layout-flex">
<figure class="wp-block-image size-full is-resized"><img loading="lazy" decoding="async" width="284" height="334" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_3.png" alt="" class="wp-image-355" style="width:124px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_3.png 284w, https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_3-255x300.png 255w" sizes="auto, (max-width: 284px) 100vw, 284px" /></figure>



<p>Let&#8217;s get to my initially mentioned point that all these SQL Server tips from randoms should be questioned. Because the statement that <strong>truncate is not rollback safe</strong> is simply wrong. The path to this realization was long for me, initiated by the conversation between Vedran and Carsten. I really learned a lot left and right, which I also wanted to record here again for posterity.</p>
</div>



<p>In the end, it doesn&#8217;t matter whether it&#8217;s <strong>DDL</strong> or <strong>DML</strong>. In the general SQL Server standard, for example, it&#8217;s described as DML, most RDBMSs in dialect with DDL. We&#8217;ll just agree on <strong>DDL</strong> and that&#8217;s it. Much more exciting is what happens under the hood.</p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<p><strong>Best comment from Vedran <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f923.png" alt="🤣" class="wp-smiley" style="height: 1em; max-height: 1em;" /></strong></p>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p class="has-overpass-mono-font-family">&#8222;Or you are trying to argue with SQL Certified Master about SQL? Good luck Mr :)&#8220;</p>
</blockquote>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">Links from the text</h2>


https://learn.microsoft.com/en-us/sql/t-sql/statements/statements?view=sql-server-ver17#data-definition-language

https://learn.microsoft.com/en-us/sql/t-sql/statements/truncate-table-transact-sql?view=sql-server-ver17

https://blog.sqlauthority.com/2010/03/04/sql-server-rollback-truncate-command-in-transaction/

https://www.sqlshack.com/the-internals-of-sql-truncate-and-sql-delete-statements/

https://www.sqlshack.com/how-to-continuously-read-transaction-log-file-data-directly-in-a-sql-server-database-with-fn_dblog-and-fn_dump_dblog/

https://en.wikipedia.org/wiki/Truncate_(SQL)

https://learn.microsoft.com/en-us/sql/relational-databases/pages-and-extents-architecture-guide?view=sql-server-ver17


<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">Full Conversation</h2>



<details class="wp-block-details is-layout-flow wp-block-details-is-layout-flow"><summary>Expand</summary>
<p><strong>Vedran Kesegic</strong></p>



<p>In sql server, truncate can be rolled back (if encapsulated within transaction). It is a logged operation.</p>



<p><strong>Carsten Saastamoinen-Jakobsen</strong></p>



<p>With DELETE we can specify the WHERE clause and some or all rows can be removed. All deleted rows are logged with BEFORE IMAGE/the data values. With TRUNCATE we can delete all rows or only the rows in one or more partitions, if we have a partition table. The only information logged is the pageid. The physical removing/releasing to unused pages happens in a background task after COMMIT. No data is stored in the log, so less and faster! And it is not possible to restore to point in time because no data is logged. The pageid in the log can already have being &#8218;released&#8216; to unused pages and therefore already used to data from another object &#8211; table/index. Before COMMIT the pages is still allocated to the table and still with data. If no explicit transaction is specified the transaction is already committed!</p>



<p><strong>Vedran Kesegic</strong></p>



<p>&#8222;No data is stored in the log&#8220; and &#8222;not possible to restore to point in time&#8220; is both incorrect for TRUNCATE. The difference is, truncate logs only deallocation of pages, which is multiple orders of magnitude less (basically insignifficant) compared to DELETE which logs data that is deleted. That is why truncate is &#8222;instant&#8220;. If you delete eg. 10GB of data, you wait for 10GB to be written to transaction log, which takes time. Truncate logs maybe 1KB for deallocation info, which is instant. If you do not want to delete all rows, you can still use TRUNCATE by storing survivor rows aside before truncating, and inserting them back after truncate. Even better, you can use SWITCH into some dummy table (even with not-partitioned tables) to remove rows quickly, and then insert survivor rows back from that dummy into original table. That is the fastest way to delete when a smaller percent of rows needs to survive.</p>



<p><strong>Vedran Kesegic</strong></p>



<p>Obvious difference is that truncate removes all rows from table (&#8222;truncate table&#8220;) or from partition(s), when using &#8222;trancate table xy with partitions (&#8230;)&#8220;. You cannot choose which rows to delete, they all are gone. With delete you can have a WHERE clause to choose which rows to delete. Delete is more appropriate if you need to delete a small percentage of table rows (or partition rows).</p>



<p><strong>Vedran Kesegic</strong></p>



<p>Truncate is a metadata operation, requires a different lock (SCH_M, schema modification lock). If also has some restrictins regarding foreign keys etc. Check the documentation for the list of limitations: https://learn.microsoft.com/en-us/sql/t-sql/statements/truncate-table-transact-sql</p>



<p><strong>Carsten Saastamoinen-Jakobsen</strong></p>



<p>It is not an argument to call TRUNCATE a DDL statement because a certain lock is used! When statements are categorized, it must be based on what happens and not how a given database system solves the task. Data is deleted, the structure is preserved and only some data may be deleted. When one or more partitions are deleted, the condition is also well-known, but just limited. In previous versions of database systems &#8211; and perhaps still &#8211; a TRUNCATE was performed by a DROP TABLE + CREATE TABLE. Hence DDL rights. It just shows that it is problematic to categorize based on how it is performed and not based on what actually happens. If we look at the documentation for SQL Server, it is not mentioned that TRUNCATE is a DDL. We may have to recognize that a previous error categorization should not continue to call TRUNCATE a DDL, but recognize that it is a DML. We will also be free from erroneous claims about only deleting all rows, that space is not released, that the transaction cannot be rolled back, &#8230;. A DELETE operation without a user-defined transaction cannot be rolled back either! But DELETE is DML and not DDL. TRUNCATE is &#8218;created&#8216; for having af fast way of deleting rows, and extended for only deleting some of the rows &#8211; DML!!!</p>



<p><strong>Vedran Kesegic</strong></p>



<p>No, the SCH-M lock (the &#8222;king of all locks&#8220;) is not the sole reason why TRUNCATE is DDL. DML writes data (rows in user tables), while DDL writes metadata (rows in system tables, data that describes other/user objects). Truncate happens to write metadata (system tables) and thus is classified as DDL. See the doc: https://learn.microsoft.com/en-us/previous-versions/sql/sql-server-2012/ff848799(v=sql.110)</p>



<p><strong>Carsten Saastamoinen-Jakobsen</strong></p>



<p>Can you tell which data is written to the system tables that is not written when removing rows by DELETE? The only additional information is, that with TRUNCATE of all rows without specifying partitions is, that the &#8218;identity&#8216; is reset to the initial values. Removing some or all rows with specifying partitions, this value is not reset to the initial value. Empty page are handled in the same way as empty page after DELETE! All indexes are preserved, the table definition is NOT changed, &#8230;.</p>



<p><strong>Vedran Kesegic</strong></p>



<p>Better than that &#8211; you can see for yourself! The exact difference and details are present in transaction log records after you do TRUNCATE and compare that to log records that DELETE generates. Use fn_dblog() function, but only on the test system because it is undocumented and might have an impact.</p>



<p>Or you are trying to say that Microsoft is wrong in documentation? If so, report it to MS so doc can be fixed. But in this case there is nothing to fix.</p>



<p>Or you are trying to argue with SQL Certified Master about SQL? Good luck Mr <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f642.png" alt="🙂" class="wp-smiley" style="height: 1em; max-height: 1em;" /></p>



<p><strong>Carsten Saastamoinen-Jakobsen</strong></p>



<p>I say that MS is right in the documentations because they are not telling that TRUNCATE is as DDL, because they know that TRUNCATE is a DML!. Of course there is less in the log &#8211; it is the idea wit TRUNCATE!!!!!!!! Just te functionality and te reason for implementing TRUNCATE as an alternative to DELETE. But the log is NOT the system tables!!!!! So please tell which system tables are updated as they are updated when CRAETE, ALTER and DROP but NOT updated when INSERT, DELETE or UPDATE.</p>



<p><strong>Vedran Kesegic</strong></p>



<p>TRUNCATE modifies internal system tables sys.sysallocunits and sys.sysrowsets. DELETE modifies indexes and stats &#8211; see attached pictures. You could do that yourself using fn_dblog as instructed, but I did that for you so you don&#8217;t have to. If you want to know more about deep SQL Server internals you can attend one of my trainings, or trainings of other SQL Masters like Brent Ozar, Paul Randal (sqlskills), etc.</p>



<p><strong>Carsten Saastamoinen-Jakobsen</strong></p>



<p>Of course they are modified as INSERT DELETE and UPDATE modify this informations. But INSERT, DELETE and UPDATE are still only DML and not DDL. And I am not impressed of your job by showing me this information. If you was showing the same for the other DML statements it would be fine. So maybe you should modify your course material and also shows the same for the other DML statements!!!!</p>
</details>



<p></p><p>The post <a href="https://dbavonnebenan.de/sql-server-truncate-vs-delete-dml-or-ddl-en/">TRUNCATE TABLE – DDL vs DML – deep dive in operations and practical tips</a> first appeared on <a href="https://dbavonnebenan.de">DBA von Nebenan</a>.</p>]]></content:encoded>
					
					<wfw:commentRss>https://dbavonnebenan.de/sql-server-truncate-vs-delete-dml-or-ddl-en/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>TRUNCATE TABLE Operationen im Detail &#8211; ROLLBACK ist doch möglich</title>
		<link>https://dbavonnebenan.de/sql-server-truncate-vs-delete-dml-or-ddl-de/</link>
					<comments>https://dbavonnebenan.de/sql-server-truncate-vs-delete-dml-or-ddl-de/#respond</comments>
		
		<dc:creator><![CDATA[Gabriel]]></dc:creator>
		<pubDate>Mon, 26 May 2025 08:16:15 +0000</pubDate>
				<category><![CDATA[Sql Server]]></category>
		<category><![CDATA[TSql]]></category>
		<category><![CDATA[DE]]></category>
		<category><![CDATA[PERFORMANCE]]></category>
		<category><![CDATA[ROLLBACK]]></category>
		<category><![CDATA[TABLE]]></category>
		<category><![CDATA[TRANSACTION]]></category>
		<category><![CDATA[TRUNCATE]]></category>
		<guid isPermaLink="false">https://dbavonnebenan.de/?p=428</guid>

					<description><![CDATA[<p>Kürzlich bin ich über einen LinkedIn-Post zu TRUNCATE vs. DELETE gestolpert, wie so viele andere auch. Typischerweise posten zufällige Mitglieder einfache SQL Server Tipps, schön verpackt in Grafiken und allgemein gehalten. Obwohl das für Anfänger absolut legitim ist &#8211; da diese einfachen Einstiegspunkte vielen Neulingen helfen können zu verstehen und einen großartigen Startpunkt bieten &#8211; [&#8230;]</p>
<p>The post <a href="https://dbavonnebenan.de/sql-server-truncate-vs-delete-dml-or-ddl-de/">TRUNCATE TABLE Operationen im Detail – ROLLBACK ist doch möglich</a> first appeared on <a href="https://dbavonnebenan.de">DBA von Nebenan</a>.</p>]]></description>
										<content:encoded><![CDATA[<p>Kürzlich bin ich über einen LinkedIn-Post zu TRUNCATE vs. DELETE gestolpert, wie so viele andere auch. Typischerweise posten zufällige Mitglieder einfache SQL Server Tipps, schön verpackt in Grafiken und allgemein gehalten. Obwohl das für Anfänger absolut legitim ist &#8211; da<strong> </strong>diese einfachen Einstiegspunkte vielen Neulingen helfen können zu verstehen und einen großartigen Startpunkt bieten &#8211; ist die Tatsache, dass dieselben Themen dann viral über verschiedene Accounts gehen, klar eine ganz andere Sache.</p>



<p>Wie auch immer. Wenn du beim Thema schon etwas weiter bist und bereits weißt, dass die Aussprache nicht &#8222;SQL&#8220; sondern &#8222;Sequel&#8220; ist <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f60a.png" alt="😊" class="wp-smiley" style="height: 1em; max-height: 1em;" />, solltest du diese Posts grundsätzlich hinterfragen, da sie meist verallgemeinern.</p>



<p>So war es auch bei einem Post über die Top 3 SQL-Fehler, die &#8222;jeder&#8220; macht. Und hier kommen wir zum Punkt &#8211; TRUNCATE TABLE und seine Besonderheiten.</p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<figure class="wp-block-image aligncenter size-full"><a href="https://www.linkedin.com/posts/samiul-fahim_sql-dataengineering-techtips-activity-7328642722653786112-OBfj?utm_source=share&amp;utm_medium=member_desktop&amp;rcm=ACoAACt1yIMBIZu3k0gdHLx7c_luIAnHYYBMtuc"><img loading="lazy" decoding="async" width="391" height="158" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-26_09-15.png" alt="TRUNCATE kann nicht zurückgerollt werden" class="wp-image-324" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-26_09-15.png 391w, https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-26_09-15-300x121.png 300w" sizes="auto, (max-width: 391px) 100vw, 391px" /></a></figure>



<p class="has-text-align-center"><a href="https://www.linkedin.com/posts/samiul-fahim_sql-dataengineering-techtips-activity-7328642722653786112-OBfj?utm_source=share&amp;utm_medium=member_desktop&amp;rcm=ACoAACt1yIMBIZu3k0gdHLx7c_luIAnHYYBMtuc" target="_blank" rel="noreferrer noopener">LinkedIn</a></p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<p>Dasselbe passierte mit <a href="https://www.linkedin.com/in/vedrankesegic/">Vedran Kesegic</a>, der den Hinweis gab, dass TRUNCATE TABLE anscheinend rollback-fähig ist, wenn es in einer Transaktion gekapselt wird. Dann sprang <a href="https://www.linkedin.com/in/saastamoinen/">Carsten Saastamoinen-Jakobsen</a> ein und die Sache wurde wild. So wild, dass mich das ganze Thema dermaßen gepackt hat, dass ich einen Deep Dive gemacht habe. Außerdem fiel in diesem Austausch so viel Wissen ab, dass es schade wäre, wenn es in einem Kommentarbereich verschwindet.</p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">Was dich hier erwartet &#8211; TRUNCATE TABLE im Detail</h2>



<p>Ich habe den Austausch verglichen und mir die Kernaussagen angeschaut. Ich habe auch separat mit beiden etwas mehr getextet, und Carsten gab mir Auszüge aus einer seiner Publikationen, wo das Thema technisch behandelt wird.</p>



<p>Technisch gesehen sind sich beide einig, auch wenn sie es nicht sagen <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f923.png" alt="🤣" class="wp-smiley" style="height: 1em; max-height: 1em;" />. Aber was am Ende wirklich wild war, war die Frage zum TRUNCATE TABLE Befehl:</p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<figure class="wp-block-image aligncenter size-full"><img loading="lazy" decoding="async" width="595" height="159" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/ddl_dml.png" alt="DML oder DDL - TRUNCATE TABLE Klassifizierung" class="wp-image-326" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/ddl_dml.png 595w, https://dbavonnebenan.de/wp-content/uploads/2025/05/ddl_dml-300x80.png 300w" sizes="auto, (max-width: 595px) 100vw, 595px" /></figure>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<div class="wp-block-columns is-layout-flex wp-container-core-columns-is-layout-28f84493 wp-block-columns-is-layout-flex">
<div class="wp-block-column is-layout-flow wp-block-column-is-layout-flow" style="flex-basis:100%">
<div class="wp-block-group is-nowrap is-layout-flex wp-container-core-group-is-layout-6c531013 wp-block-group-is-layout-flex">
<figure class="wp-block-image size-full"><img loading="lazy" decoding="async" width="246" height="470" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_standing_1.png" alt="" class="wp-image-330" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_standing_1.png 246w, https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_standing_1-157x300.png 157w" sizes="auto, (max-width: 246px) 100vw, 246px" /></figure>



<p>Ich kann es vorwegnehmen. Diese Frage zur TRUNCATE TABLE Klassifizierung ist eine philosophische. Ich konnte mir basierend auf meiner Recherche eine Meinung bilden, mehr dazu später, aber grundsätzlich liegt es an jedem selbst zu entscheiden.</p>
</div>
</div>
</div>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">Die Diskussion &#8211; TRUNCATE TABLE als DML oder DDL?</h2>



<h3 class="wp-block-heading">Vedran Kesegic</h3>



<p><strong>Kernposition: TRUNCATE TABLE ist DDL</strong></p>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p>&#8222;TRUNCATE modifies internal system tables sys.sysallocunits and sys.sysrowsets. DELETE modifies indexes and stats&#8220;</p>
</blockquote>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p>&#8222;DML writes data (rows in user tables), while DDL writes metadata (rows in system tables, data that describes other/user objects). Truncate happens to write metadata (system tables) and thus is classified as DDL.&#8220;</p>
</blockquote>



<p><strong>Technische Argumente für TRUNCATE TABLE als DDL</strong></p>



<ul class="wp-block-list">
<li>Verwendet SCH-M (Schema Modification) Lock</li>



<li>Modifiziert System-Metadaten-Tabellen</li>



<li>Microsoft-Dokumentation klassifiziert es als DDL</li>



<li>Loggt Page-Deallocations, nicht Row-Deletions</li>
</ul>



<p><strong>Kernaussage zum TRUNCATE TABLE Befehl</strong></p>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p>&#8222;TRUNCATE modifies internal system tables sys.sysallocunits and sys.sysrowsets. DELETE modifies indexes and stats &#8211; see attached&#8220;</p>
</blockquote>



<div style="height:50px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">Carsten Saastamoinen-Jakobsen</h3>



<p><strong>Kernposition: TRUNCATE TABLE ist DML</strong></p>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p>&#8222;It is not an argument to call TRUNCATE a DDL statement because a certain lock is used! When statements are categorized, it must be based on what happens and not how a given database system solves the task.&#8220;</p>
</blockquote>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p>&#8222;We may have to recognize that a previous error categorization should not continue to call TRUNCATE a DDL, but recognize that it is a DML.&#8220;</p>
</blockquote>



<p><strong>Logische Argumente für TRUNCATE TABLE als DML</strong></p>



<ul class="wp-block-list">
<li>Daten werden gelöscht, Struktur bleibt erhalten</li>



<li>Keine Änderungen an der Tabellendefinition</li>



<li>Folgt semantischen Regeln der Datenmanipulation</li>



<li>Kann nur bestimmte Partitionen löschen (nicht alle Daten)</li>
</ul>



<p><strong>Kernaussage zur TRUNCATE TABLE Klassifizierung</strong></p>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p>&#8222;I say that MS is right in the documentations because they are not telling that TRUNCATE is as DDL, because they know that TRUNCATE is a DML!&#8220;</p>



<p><mark style="background-color:rgba(0, 0, 0, 0)" class="has-inline-color has-denary-color"><em>ANMERKUNG Gabriel <br></em></mark><em>Ich denke, das ist wirklich etwas fragwürdig, weil die Dokumentation absolut klar</em> <em>zu diesem Punkt ist</em> &#8211;&gt; <strong><em>DDL </em></strong>(<a href="https://learn.microsoft.com/en-us/sql/t-sql/statements/statements?view=sql-server-ver17#data-definition-language" target="_blank" rel="noreferrer noopener">microsoft.com</a>)<br></p>
</blockquote>



<p><strong>Zitat aus der Publikation zum TRUNCATE TABLE</strong></p>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p>&#8222;As of version 2016, it is possible to delete only some of the data in a table with TRUNCATE, so it is even more cor­rect to call it manipulation than definition, because it is changing some of the data but not changing the structure.&#8220;</p>
</blockquote>



<div style="height:50px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">Einigungspunkte bei TRUNCATE TABLE</h3>



<ul class="wp-block-list">
<li>TRUNCATE TABLE kann in einer Transaktion zurückgerollt werden
<ul class="wp-block-list">
<li>allgemeine Demo bei <a href="https://blog.sqlauthority.com/2010/03/04/sql-server-rollback-truncate-command-in-transaction/" target="_blank" rel="noreferrer noopener">sqlauthority.com</a></li>



<li>detaillierte Demo <a href="https://www.sqlshack.com/the-internals-of-sql-truncate-and-sql-delete-statements/" target="_blank" rel="noreferrer noopener">sqlshack.com</a></li>
</ul>
</li>



<li>TRUNCATE TABLE ist eine geloggte Operation</li>



<li>TRUNCATE TABLE ist schneller als DELETE bei großen Datensätzen</li>



<li>Beide verstehen die technische Implementierung</li>
</ul>



<div style="height:50px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">Hauptstreitpunkt bei TRUNCATE TABLE</h3>



<ul class="wp-block-list">
<li><strong>Klassifizierungskriterien für TRUNCATE TABLE</strong>
<ul class="wp-block-list">
<li>Implementierungsdetails</li>
</ul>
</li>



<li><strong>Autorität</strong>
<ul class="wp-block-list">
<li>Vendor-Dokumentation (Vedran) vs. Logische Begründung (Carsten)</li>
</ul>
</li>



<li><strong>Perspektive zum TRUNCATE TABLE Befehl</strong>
<ul class="wp-block-list">
<li>Wie es intern funktioniert (Vedran) vs. Was es funktional macht (Carsten)</li>
</ul>
</li>
</ul>



<div style="height:50px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">Meine Einschätzung zu TRUNCATE TABLE</h3>



<p>Wie du siehst, passt es technisch, aber wie betrachtest du das Ganze? Vielleicht ist es gut, dass ich nicht mit so viel Wissen vorab beladen bin wie beide, denn aus meiner Sicht ist es ziemlich klar: <strong>TRUNCATE TABLE ist DDL!!</strong></p>



<div class="wp-block-group is-nowrap is-layout-flex wp-container-core-group-is-layout-6c531013 wp-block-group-is-layout-flex">
<figure class="wp-block-image size-full is-resized"><img loading="lazy" decoding="async" width="292" height="342" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_1.png" alt="" class="wp-image-344" style="width:98px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_1.png 292w, https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_1-256x300.png 256w" sizes="auto, (max-width: 292px) 100vw, 292px" /></figure>



<p>Entgegen Carstens Aussage, dass keine Systemtabellen geändert werden, ist Vedrans Aussage absolut korrekt. Sieh warum&#8230; </p>
</div>



<p>&#8230; und natürlich weil Microsoft sagt, dass TRUNCATE TABLE DDL ist <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f60a.png" alt="😊" class="wp-smiley" style="height: 1em; max-height: 1em;" /> ( <a href="https://learn.microsoft.com/en-us/sql/t-sql/statements/statements?view=sql-server-ver17#data-definition-language" target="_blank" rel="noreferrer noopener">microsoft.com</a> )</p>



<p>Und Partitionierung ist DML… ich weiß nicht, auch hier sind es zumindest Systemspalten, die die Partitionierung definieren.</p>



<p></p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">Wie TRUNCATE TABLE funktioniert &#8211; Schritt für Schritt</h2>



<p>Ich würde gerne sagen, dass ich es einfach beschreibe, aber das ist einfach meine normale Beschreibung. Das Thema TRUNCATE TABLE ist wirklich komplex <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/270c.png" alt="✌" class="wp-smiley" style="height: 1em; max-height: 1em;" /></p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">1 &#8211; TRUNCATE TABLE Befehl initiiert</h3>


<div class="wp-block-syntaxhighlighter-code "><pre class="brush: sql; title: ; notranslate">
TRUNCATE TABLE dbo.MyTable;
</pre></div>


<ul class="wp-block-list">
<li><strong>Erste Aktion</strong>: Akquiriert <strong>SCH-M (Schema Modification)</strong> Lock auf gesamte Tabelle</li>



<li><strong>Effekt</strong>: Tabelle ist komplett gesperrt &#8211; keine Reads, keine Writes, nichts</li>
</ul>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">2 &#8211; Extents identifizieren bei TRUNCATE TABLE</h3>



<ul class="wp-block-list">
<li>DEEP DIVE Extents &#8211;&gt; <a href="https://learn.microsoft.com/en-us/sql/relational-databases/pages-and-extents-architecture-guide?view=sql-server-ver17" target="_blank" rel="noreferrer noopener">microsoft.com</a></li>



<li>SQL Server liest IAM-Pages</li>



<li>Listet alle Extents auf, die zur Tabelle gehören</li>



<li><strong>Hält weiterhin</strong>: SCH-M Lock auf Tabelle</li>
</ul>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">3 &#8211; Einzelne Extents bei TRUNCATE TABLE sperren</h3>



<ul class="wp-block-list">
<li>Platziert <strong>X (Exclusive)</strong> Locks auf jeden Extent</li>



<li>Das sind reguläre Data-Locks, KEINE Schema-Locks</li>



<li><strong>Hält jetzt</strong>:
<ul class="wp-block-list">
<li>SCH-M Lock auf Tabelle (verhindert Zugriff)</li>



<li>X Locks auf Extents (markiert sie für Deallocation)</li>
</ul>
</li>
</ul>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">4 &#8211; TRUNCATE TABLE Operation loggen</h3>



<ul class="wp-block-list">
<li>Schreibt ins Transaction-Log: &#8222;Deallocate extents 8:8, 8:16, etc.&#8220;</li>



<li><strong>Hält weiterhin</strong>: Beide Lock-Typen</li>
</ul>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">5a &#8211; Bei COMMIT der TRUNCATE TABLE Operation</h3>



<ul class="wp-block-list">
<li>Aktualisiert IAM &#8211; markiert Extents als frei</li>



<li><strong>Gibt frei</strong> alle X Locks auf Extents</li>



<li><strong>Gibt frei</strong> SCH-M Lock auf Tabelle</li>



<li>Tabelle ist wieder zugänglich</li>
</ul>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">5b &#8211; Bei ROLLBACK der TRUNCATE TABLE Operation</h3>



<ul class="wp-block-list">
<li>Verwirft Deallocation-Marker</li>



<li><strong>Gibt frei</strong> alle X Locks auf Extents</li>



<li><strong>Gibt frei</strong> SCH-M Lock auf Tabelle</li>



<li>Alles bleibt unverändert</li>
</ul>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">Warum zwei verschiedene Lock-Typen bei TRUNCATE TABLE?</h3>



<p><strong>SCH-M Lock:</strong></p>



<ul class="wp-block-list">
<li>&#8222;Niemand rührt diese Tabelle ÜBERHAUPT an&#8220;</li>



<li>Typisch für DDL-Operationen</li>



<li>Verhindert sogar Metadaten-Abfragen</li>
</ul>



<p><strong>X Locks auf Extents:</strong></p>



<ul class="wp-block-list">
<li>&#8222;Diese spezifischen Datenblöcke werden modifiziert&#8220;</li>



<li>Typisch für Daten-Operationen</li>



<li>Stellt sicher, dass Extents nicht von anderen Sessions wiederverwendet werden bis commit/rollback</li>
</ul>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">Meine bescheidene Meinung zu TRUNCATE TABLE</h3>



<p>Mit diesem Wissen ist der Fall eigentlich klar für mich, aber wie siehst du es? Tatsächlich werden hier keine Benutzerdaten manipuliert; stattdessen werden System-Level-Zuweisungen angepasst. Folglich erlaube ich mir, trotz meines neuen Halbwissens, den folgenden Vergleich zu ziehen:</p>



<div class="wp-block-group is-nowrap is-layout-flex wp-container-core-group-is-layout-6c531013 wp-block-group-is-layout-flex">
<figure class="wp-block-image size-full is-resized"><img loading="lazy" decoding="async" width="307" height="328" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_2.png" alt="" class="wp-image-345" style="width:131px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_2.png 307w, https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_2-281x300.png 281w" sizes="auto, (max-width: 307px) 100vw, 307px" /></figure>



<p>&#8222;Ich sage auch nicht, dass ich Text in einem Word-Dokument lösche, wenn ich die Festplatte schnell formatiere und nur die Block-Allocation lösche.&#8220;</p>
</div>



<p><br>Ja, ich denke das ist ein schöner Vergleich&#8230; Fall abgeschlossen, jetzt ein paar technisch interessante Aspekte und Lösungen zum TRUNCATE TABLE Befehl</p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">Technischer Teil, Hacks und TRUNCATE TABLE How-To</h2>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<figure class="wp-block-image aligncenter size-full is-resized"><img loading="lazy" decoding="async" width="643" height="388" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/meme_trunc.jpg" alt="" class="wp-image-348" style="width:325px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/meme_trunc.jpg 643w, https://dbavonnebenan.de/wp-content/uploads/2025/05/meme_trunc-300x181.jpg 300w" sizes="auto, (max-width: 643px) 100vw, 643px" /></figure>



<p class="has-text-align-center has-medium-font-size">&#8230; praktisch, nicht technisch <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f609.png" alt="😉" class="wp-smiley" style="height: 1em; max-height: 1em;" /></p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">TRUNCATE TABLE mit Survivor Table (Temporary Table)</h3>



<p>Daten, die noch benötigt werden, werden einfach in eine temporäre Tabelle verschoben und nach TRUNCATE TABLE wieder abgerufen</p>


<div class="wp-block-syntaxhighlighter-code "><pre class="brush: sql; title: ; notranslate">
-- Step 1: Create temp table with rows to keep
	SELECT * INTO #SurvivorRows 
	FROM 
		MyTable 
	WHERE
		DateColumn &gt;= &#039;2024-01-01&#039;

-- Step 2: Truncate the entire table
	TRUNCATE TABLE MyTable

-- Step 3: Re-insert survivor rows
	INSERT INTO MyTable
	SELECT * FROM #SurvivorRows

-- Step 4: Clean up
	DROP TABLE #SurvivorRows
</pre></div>


<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">TRUNCATE TABLE mit PARTITION Option (2016++)</h3>



<p>Seit SQL Server 2016 (jede Edition) ist es möglich, Partitionen oder Bereiche von Partitionen bei TRUNCATE TABLE zu spezifizieren</p>


<div class="wp-block-syntaxhighlighter-code "><pre class="brush: sql; title: ; notranslate">
-- Truncate specific partitions
	TRUNCATE TABLE MyPartitionedTable 
	     WITH (PARTITIONS (2, 4 TO 6))

-- Check partition status
	SELECT 
		 partition_number
		,rows 
	FROM
		sys.partitions 
	WHERE 
		OBJECT_NAME(OBJECT_ID) = &#039;MyPartitionedTable&#039;
</pre></div>


<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">TRUNCATE TABLE ist immer eine Transaktion&#8230;</h3>



<p>&#8230;. und muss beim RESTORE berücksichtigt werden. Carsten hat das ganze sehr schön in seiner Publikation illustriert. Ich beschreibe es nur, das muss an dieser Stelle reichen <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f61d.png" alt="😝" class="wp-smiley" style="height: 1em; max-height: 1em;" /></p>



<p>Obwohl wir nur <strong>TRUNCATE TABLE</strong> schreiben, verwendet SQL Server trotzdem <strong>BEGIN</strong> und <strong>COMMIT TRANSACTION</strong>, was natürlich im TLog aufgezeichnet wird. Entsprechend bleiben bei einer POINT IN TIME RECOVERY die Daten in der Tabelle verfügbar, bis das TLog mit der entsprechenden TRANSACTION wiederhergestellt wird, oder die Zeit der TRUNCATE TABLE Transaktion überschritten wurde.</p>



<p>Du kannst also auch die Daten wiederherstellen <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f44c.png" alt="👌" class="wp-smiley" style="height: 1em; max-height: 1em;" /></p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h3 class="wp-block-heading">Transaction Log File Daten kontinuierlich lesen mit fn_dblog und fn_dump_dblog</h3>



<p>Vedran schrieb in einem seiner Posts, dass es eine undokumentierte Funktion gibt, wo du Log-Records zur Laufzeit lesen kannst. Es gibt auch einen sehr coolen Artikel dazu auf sqlshack.com</p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<p class="has-text-align-center"><em><mark style="background-color:rgba(0, 0, 0, 0)" class="has-inline-color has-septenary-color">NICHT empfohlen für Produktion, sondern nur für Test / Bildung.</mark></em></p>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>


https://www.sqlshack.com/how-to-continuously-read-transaction-log-file-data-directly-in-a-sql-server-database-with-fn_dblog-and-fn_dump_dblog/


<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">Fazit zu TRUNCATE TABLE</h2>



<div class="wp-block-group is-nowrap is-layout-flex wp-container-core-group-is-layout-6c531013 wp-block-group-is-layout-flex">
<figure class="wp-block-image size-full is-resized"><img loading="lazy" decoding="async" width="284" height="334" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_3.png" alt="" class="wp-image-355" style="width:124px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_3.png 284w, https://dbavonnebenan.de/wp-content/uploads/2025/05/emotes_3-255x300.png 255w" sizes="auto, (max-width: 284px) 100vw, 284px" /></figure>



<p>Kommen wir zu meinem eingangs erwähnten Punkt, dass all diese SQL Server Tipps von randoms hinterfragt werden sollten. Denn die Aussage, dass <strong>TRUNCATE TABLE nicht rollback-sicher</strong> ist, ist schlichtweg falsch. Der Weg zu dieser Erkenntnis war für mich lang, eingeleitet durch die Unterhaltung zwischen Vedran und Carsten. Ich habe wirklich viel links und rechts dazugelernt, was ich auch hier nochmal für die Nachwelt festhalten wollte.</p>
</div>



<p>Am Ende ist es egal, ob TRUNCATE TABLE <strong>DDL</strong> oder <strong>DML</strong> ist. Im allgemeinen SQL Server Standard wird es beispielsweise als DML beschrieben, die meisten RDBMSs im Dialekt mit DDL. Wir einigen uns einfach auf <strong>DDL</strong> und gut ist. Viel spannender ist, was unter der Haube beim TRUNCATE TABLE Befehl passiert.</p>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<p><strong>Bester Kommentar von Vedran <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f923.png" alt="🤣" class="wp-smiley" style="height: 1em; max-height: 1em;" /></strong></p>



<blockquote class="wp-block-quote is-layout-flow wp-block-quote-is-layout-flow">
<p class="has-overpass-mono-font-family">&#8222;Or you are trying to argue with SQL Certified Master about SQL? Good luck Mr :)&#8220;</p>
</blockquote>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">Links aus dem Text</h2>


https://learn.microsoft.com/en-us/sql/t-sql/statements/statements?view=sql-server-ver17#data-definition-language

https://learn.microsoft.com/en-us/sql/t-sql/statements/truncate-table-transact-sql?view=sql-server-ver17

https://blog.sqlauthority.com/2010/03/04/sql-server-rollback-truncate-command-in-transaction/

https://www.sqlshack.com/the-internals-of-sql-truncate-and-sql-delete-statements/

https://www.sqlshack.com/how-to-continuously-read-transaction-log-file-data-directly-in-a-sql-server-database-with-fn_dblog-and-fn_dump_dblog/

https://en.wikipedia.org/wiki/Truncate_(SQL)

https://learn.microsoft.com/en-us/sql/relational-databases/pages-and-extents-architecture-guide?view=sql-server-ver17


<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:32px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">Full Conversation</h2>



<details class="wp-block-details is-layout-flow wp-block-details-is-layout-flow"><summary>Expand</summary>
<p><strong>Vedran Kesegic</strong></p>



<p>In sql server, truncate can be rolled back (if encapsulated within transaction). It is a logged operation.</p>



<p><strong>Carsten Saastamoinen-Jakobsen</strong></p>



<p>With DELETE we can specify the WHERE clause and some or all rows can be removed. All deleted rows are logged with BEFORE IMAGE/the data values. With TRUNCATE we can delete all rows or only the rows in one or more partitions, if we have a partition table. The only information logged is the pageid. The physical removing/releasing to unused pages happens in a background task after COMMIT. No data is stored in the log, so less and faster! And it is not possible to restore to point in time because no data is logged. The pageid in the log can already have being &#8218;released&#8216; to unused pages and therefore already used to data from another object &#8211; table/index. Before COMMIT the pages is still allocated to the table and still with data. If no explicit transaction is specified the transaction is already committed!</p>



<p><strong>Vedran Kesegic</strong></p>



<p>&#8222;No data is stored in the log&#8220; and &#8222;not possible to restore to point in time&#8220; is both incorrect for TRUNCATE. The difference is, truncate logs only deallocation of pages, which is multiple orders of magnitude less (basically insignifficant) compared to DELETE which logs data that is deleted. That is why truncate is &#8222;instant&#8220;. If you delete eg. 10GB of data, you wait for 10GB to be written to transaction log, which takes time. Truncate logs maybe 1KB for deallocation info, which is instant. If you do not want to delete all rows, you can still use TRUNCATE by storing survivor rows aside before truncating, and inserting them back after truncate. Even better, you can use SWITCH into some dummy table (even with not-partitioned tables) to remove rows quickly, and then insert survivor rows back from that dummy into original table. That is the fastest way to delete when a smaller percent of rows needs to survive.</p>



<p><strong>Vedran Kesegic</strong></p>



<p>Obvious difference is that truncate removes all rows from table (&#8222;truncate table&#8220;) or from partition(s), when using &#8222;trancate table xy with partitions (&#8230;)&#8220;. You cannot choose which rows to delete, they all are gone. With delete you can have a WHERE clause to choose which rows to delete. Delete is more appropriate if you need to delete a small percentage of table rows (or partition rows).</p>



<p><strong>Vedran Kesegic</strong></p>



<p>Truncate is a metadata operation, requires a different lock (SCH_M, schema modification lock). If also has some restrictins regarding foreign keys etc. Check the documentation for the list of limitations: https://learn.microsoft.com/en-us/sql/t-sql/statements/truncate-table-transact-sql</p>



<p><strong>Carsten Saastamoinen-Jakobsen</strong></p>



<p>It is not an argument to call TRUNCATE a DDL statement because a certain lock is used! When statements are categorized, it must be based on what happens and not how a given database system solves the task. Data is deleted, the structure is preserved and only some data may be deleted. When one or more partitions are deleted, the condition is also well-known, but just limited. In previous versions of database systems &#8211; and perhaps still &#8211; a TRUNCATE was performed by a DROP TABLE + CREATE TABLE. Hence DDL rights. It just shows that it is problematic to categorize based on how it is performed and not based on what actually happens. If we look at the documentation for SQL Server, it is not mentioned that TRUNCATE is a DDL. We may have to recognize that a previous error categorization should not continue to call TRUNCATE a DDL, but recognize that it is a DML. We will also be free from erroneous claims about only deleting all rows, that space is not released, that the transaction cannot be rolled back, &#8230;. A DELETE operation without a user-defined transaction cannot be rolled back either! But DELETE is DML and not DDL. TRUNCATE is &#8218;created&#8216; for having af fast way of deleting rows, and extended for only deleting some of the rows &#8211; DML!!!</p>



<p><strong>Vedran Kesegic</strong></p>



<p>No, the SCH-M lock (the &#8222;king of all locks&#8220;) is not the sole reason why TRUNCATE is DDL. DML writes data (rows in user tables), while DDL writes metadata (rows in system tables, data that describes other/user objects). Truncate happens to write metadata (system tables) and thus is classified as DDL. See the doc: https://learn.microsoft.com/en-us/previous-versions/sql/sql-server-2012/ff848799(v=sql.110)</p>



<p><strong>Carsten Saastamoinen-Jakobsen</strong></p>



<p>Can you tell which data is written to the system tables that is not written when removing rows by DELETE? The only additional information is, that with TRUNCATE of all rows without specifying partitions is, that the &#8218;identity&#8216; is reset to the initial values. Removing some or all rows with specifying partitions, this value is not reset to the initial value. Empty page are handled in the same way as empty page after DELETE! All indexes are preserved, the table definition is NOT changed, &#8230;.</p>



<p><strong>Vedran Kesegic</strong></p>



<p>Better than that &#8211; you can see for yourself! The exact difference and details are present in transaction log records after you do TRUNCATE and compare that to log records that DELETE generates. Use fn_dblog() function, but only on the test system because it is undocumented and might have an impact.</p>



<p>Or you are trying to say that Microsoft is wrong in documentation? If so, report it to MS so doc can be fixed. But in this case there is nothing to fix.</p>



<p>Or you are trying to argue with SQL Certified Master about SQL? Good luck Mr <img src="https://s.w.org/images/core/emoji/17.0.2/72x72/1f642.png" alt="🙂" class="wp-smiley" style="height: 1em; max-height: 1em;" /></p>



<p><strong>Carsten Saastamoinen-Jakobsen</strong></p>



<p>I say that MS is right in the documentations because they are not telling that TRUNCATE is as DDL, because they know that TRUNCATE is a DML!. Of course there is less in the log &#8211; it is the idea wit TRUNCATE!!!!!!!! Just te functionality and te reason for implementing TRUNCATE as an alternative to DELETE. But the log is NOT the system tables!!!!! So please tell which system tables are updated as they are updated when CRAETE, ALTER and DROP but NOT updated when INSERT, DELETE or UPDATE.</p>



<p><strong>Vedran Kesegic</strong></p>



<p>TRUNCATE modifies internal system tables sys.sysallocunits and sys.sysrowsets. DELETE modifies indexes and stats &#8211; see attached pictures. You could do that yourself using fn_dblog as instructed, but I did that for you so you don&#8217;t have to. If you want to know more about deep SQL Server internals you can attend one of my trainings, or trainings of other SQL Masters like Brent Ozar, Paul Randal (sqlskills), etc.</p>



<p><strong>Carsten Saastamoinen-Jakobsen</strong></p>



<p>Of course they are modified as INSERT DELETE and UPDATE modify this informations. But INSERT, DELETE and UPDATE are still only DML and not DDL. And I am not impressed of your job by showing me this information. If you was showing the same for the other DML statements it would be fine. So maybe you should modify your course material and also shows the same for the other DML statements!!!!</p>
</details>



<p></p><p>The post <a href="https://dbavonnebenan.de/sql-server-truncate-vs-delete-dml-or-ddl-de/">TRUNCATE TABLE Operationen im Detail – ROLLBACK ist doch möglich</a> first appeared on <a href="https://dbavonnebenan.de">DBA von Nebenan</a>.</p>]]></content:encoded>
					
					<wfw:commentRss>https://dbavonnebenan.de/sql-server-truncate-vs-delete-dml-or-ddl-de/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>SysAdmin Role ignores the Default User Schema &#8211; A confusing story</title>
		<link>https://dbavonnebenan.de/sql-server-sysadmin-role-can-override-the-default-schema-en/</link>
					<comments>https://dbavonnebenan.de/sql-server-sysadmin-role-can-override-the-default-schema-en/#respond</comments>
		
		<dc:creator><![CDATA[Gabriel]]></dc:creator>
		<pubDate>Thu, 08 May 2025 10:20:44 +0000</pubDate>
				<category><![CDATA[Sql Server]]></category>
		<category><![CDATA[TSql]]></category>
		<category><![CDATA[default]]></category>
		<category><![CDATA[eng]]></category>
		<category><![CDATA[schema]]></category>
		<guid isPermaLink="false">https://dbavonnebenan.de/?p=279</guid>

					<description><![CDATA[<p>Today I had another call on my desk where an error message from an application stated that an object in the Sql Server DB couldn&#8217;t be accessed. The vendor claimed &#8222;connection problem&#8220;, I said &#8222;DB or at least server problem&#8220;, customer says &#8222;everything used to work before&#8220;. The Extended Events were quickly set up, &#8222;Remote [&#8230;]</p>
<p>The post <a href="https://dbavonnebenan.de/sql-server-sysadmin-role-can-override-the-default-schema-en/">SysAdmin Role ignores the Default User Schema – A confusing story</a> first appeared on <a href="https://dbavonnebenan.de">DBA von Nebenan</a>.</p>]]></description>
										<content:encoded><![CDATA[<p>Today I had another call on my desk where an error message from an application stated that an object in the Sql Server DB couldn&#8217;t be accessed. The vendor claimed &#8222;connection problem&#8220;, I said &#8222;DB or at least server problem&#8220;, customer says &#8222;everything used to work before&#8220;.</p>



<p>The Extended Events were quickly set up, &#8222;Remote Procedure Call Starting&#8220; shows me a statement without a fully qualified table name. The target object is in the database under the db_owner schema. Welllll, that&#8217;s already questionable, but OK. Why not keep it as dbo?</p>



<p>Anyway, the behavior at this point was really incomprehensible to me since everything was configured correctly: Default schema of the user, db_owner on the DB. And the SysAdmin. I had noticed that, but well, customer&#8217;s decision. I&#8217;ll note things like this in a session, but won&#8217;t change anything to avoid endangering grown dependencies.</p>



<p>After searching for a long time, I almost gave up, referred to the vendor and ended the call. Unfortunately, I can&#8217;t let go of stuff like this, so I fired up the lab, rebuilt the environment and lo and behold, found the error. </p>



<p>The problem is the SysAdmin role. It directly maps the login to the dbo user of the database &#8211; regardless of whether the actual user exists or not, thus inheriting dbo&#8217;s default schema. Mindblowing, didn&#8217;t know that. 🙂</p>



<div style="height:13px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">SysAdmin Role InActive</h2>



<figure class="wp-block-image size-full is-resized has-custom-border"><img loading="lazy" decoding="async" width="655" height="449" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-49_1.png" alt="" class="has-border-color has-base-border-color wp-image-285" style="border-width:1px;width:641px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-49_1.png 655w, https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-49_1-300x206.png 300w" sizes="auto, (max-width: 655px) 100vw, 655px" /></figure>



<div style="height:13px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">SysAdmin Role Active</h2>



<figure class="wp-block-image size-full is-resized has-custom-border"><img loading="lazy" decoding="async" width="667" height="351" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-50.png" alt="" class="has-border-color has-base-border-color wp-image-287" style="border-width:1px;width:631px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-50.png 667w, https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-50-300x158.png 300w" sizes="auto, (max-width: 667px) 100vw, 667px" /></figure>



<figure class="wp-block-image size-large is-resized has-custom-border"><img loading="lazy" decoding="async" width="504" height="212" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-13.png" alt="" class="has-border-color has-base-border-color wp-image-281" style="border-width:1px;width:495px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-13.png 504w, https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-13-300x126.png 300w" sizes="auto, (max-width: 504px) 100vw, 504px" /></figure>



<div style="height:13px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:13px" aria-hidden="true" class="wp-block-spacer"></div>



<details class="wp-block-details is-layout-flow wp-block-details-is-layout-flow"><summary>T-Sql CODE I used for the screenshots</summary><div class="wp-block-syntaxhighlighter-code "><pre class="brush: sql; title: ; notranslate">
DECLARE @IsSysAdmin INT
SELECT @IsSysAdmin = IS_SRVROLEMEMBER(&#039;sysadmin&#039;)


SELECT SYSTEM_USER AS CurrentLoginName;

SELECT 
    &#039;Current User&#039;  AS Info,
    USER_NAME()     AS Username,
    SCHEMA_NAME()   AS CurrentSchema,
    CASE WHEN @IsSysAdmin = 1 THEN &#039;Yes&#039; ELSE &#039;No&#039; END 
                    AS IsSysAdmin


SELECT 
    &#039;Default Schema&#039;        AS Info,
    dp.name                 AS Username,
    dp.default_schema_name AS DefaultSchema
FROM 
    sys.database_principals dp
WHERE 
    dp.name = USER_NAME()


SELECT 
    &#039;DB Role Membership&#039;    AS Info,
    USER_NAME()             AS Username,
    r.name                  AS RoleName
FROM 
    sys.database_principals u
JOIN 
    sys.database_role_members m ON u.principal_id = m.member_principal_id
JOIN 
    sys.database_principals r ON m.role_principal_id = r.principal_id
WHERE 
    u.name = USER_NAME()

--------------------

-- select * from &#x5B;customSchema].&#x5B;userTable] &lt;&lt;&lt;&lt;&lt; ALWAYS WORKING !!!!!

 select * from &#x5B;userTable]



</pre></div></details>



<div style="height:13px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div class="wp-block-group is-nowrap is-layout-flex wp-container-core-group-is-layout-6c531013 wp-block-group-is-layout-flex">
<figure class="wp-block-image size-full"><img loading="lazy" decoding="async" width="283" height="292" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-43.png" alt="" class="wp-image-284"/></figure>



<ul class="wp-block-list">
<li>Always use fully qualified object names, someday you&#8217;ll be glad you did. You can read this in many blogs.</li>



<li>Always work according to POLP (<a href="https://en.wikipedia.org/wiki/Principle_of_least_privilege">Principle of Least Privilege</a>), an app user shouldn&#8217;t have SysAdmin rights! 🙂</li>
</ul>
</div>



<p class="has-text-align-center"><strong>Gabriel, der DBAvonNebenan</strong></p>



<p></p><p>The post <a href="https://dbavonnebenan.de/sql-server-sysadmin-role-can-override-the-default-schema-en/">SysAdmin Role ignores the Default User Schema – A confusing story</a> first appeared on <a href="https://dbavonnebenan.de">DBA von Nebenan</a>.</p>]]></content:encoded>
					
					<wfw:commentRss>https://dbavonnebenan.de/sql-server-sysadmin-role-can-override-the-default-schema-en/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>SysAdmin Role ignoriert das Default User Schema – Eine verwirrende Fehlersuche</title>
		<link>https://dbavonnebenan.de/sql-server-sysadmin-role-can-override-the-default-schema-de/</link>
					<comments>https://dbavonnebenan.de/sql-server-sysadmin-role-can-override-the-default-schema-de/#respond</comments>
		
		<dc:creator><![CDATA[Gabriel]]></dc:creator>
		<pubDate>Thu, 08 May 2025 10:13:05 +0000</pubDate>
				<category><![CDATA[TSql]]></category>
		<category><![CDATA[DBO]]></category>
		<category><![CDATA[DE]]></category>
		<category><![CDATA[schema]]></category>
		<category><![CDATA[SYSADMIN]]></category>
		<guid isPermaLink="false">https://dbavonnebenan.de/?p=434</guid>

					<description><![CDATA[<p>Heute hatte ich wieder einen Fall auf dem Tisch, bei dem eine Fehlermeldung einer Anwendung besagte, dass auf ein Objekt in der SQL Server DB nicht zugegriffen werden kann. Der Hersteller behauptete &#8222;Verbindungsproblem&#8220;, ich sagte &#8222;DB oder zumindest Server-Problem&#8220;, Kunde sagt &#8222;früher hat alles funktioniert&#8220;. Die Extended Events waren schnell aufgesetzt, &#8222;Remote Procedure Call Starting&#8220; [&#8230;]</p>
<p>The post <a href="https://dbavonnebenan.de/sql-server-sysadmin-role-can-override-the-default-schema-de/">SysAdmin Role ignoriert das Default User Schema – Eine verwirrende Fehlersuche</a> first appeared on <a href="https://dbavonnebenan.de">DBA von Nebenan</a>.</p>]]></description>
										<content:encoded><![CDATA[<p>Heute hatte ich wieder einen Fall auf dem Tisch, bei dem eine Fehlermeldung einer Anwendung besagte, dass auf ein Objekt in der SQL Server DB nicht zugegriffen werden kann. Der Hersteller behauptete &#8222;Verbindungsproblem&#8220;, ich sagte &#8222;DB oder zumindest Server-Problem&#8220;, Kunde sagt &#8222;früher hat alles funktioniert&#8220;.</p>



<p>Die Extended Events waren schnell aufgesetzt, &#8222;Remote Procedure Call Starting&#8220; zeigt mir ein Statement ohne vollqualifizierten Tabellennamen. Das Zielobjekt liegt in der Database unter dem db_owner Schema. Najaaa, das ist schon fragwürdig, aber OK. Warum nicht als dbo belassen?</p>



<p>Jedenfalls war mir das Verhalten an diesem Punkt wirklich unverständlich, da alles korrekt konfiguriert war: Default Schema des Users, db_owner auf der DB. Und der SysAdmin. Das war mir aufgefallen, aber nun ja, Entscheidung des Kunden. Sowas notiere ich mir in einer Session, ändere aber nichts, um gewachsene Abhängigkeiten nicht zu gefährden.</p>



<p>Nach langem Suchen gab ich fast auf, verwies an den Vendor und beendete den Call. Leider kann ich solche Sachen nicht loslassen, also feuerte ich das Lab an, baute die Umgebung nach und siehe da, fand den Fehler.</p>



<p>Das Problem ist die SysAdmin-Rolle. Sie mappt den Login direkt auf den dbo-User der Database &#8211; egal ob der eigentliche User existiert oder nicht, erbt somit das Default Schema von dbo. Mindblowing, wusste ich nicht. 🙂</p>



<div style="height:13px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">SysAdmin Role inaktiv</h2>



<figure class="wp-block-image size-full is-resized has-custom-border"><img loading="lazy" decoding="async" width="655" height="449" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-49_1.png" alt="" class="has-border-color has-base-border-color wp-image-285" style="border-width:1px;width:641px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-49_1.png 655w, https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-49_1-300x206.png 300w" sizes="auto, (max-width: 655px) 100vw, 655px" /></figure>



<div style="height:13px" aria-hidden="true" class="wp-block-spacer"></div>



<h2 class="wp-block-heading">SysAdmin Role aktiv</h2>



<figure class="wp-block-image size-full is-resized has-custom-border"><img loading="lazy" decoding="async" width="667" height="351" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-50.png" alt="" class="has-border-color has-base-border-color wp-image-287" style="border-width:1px;width:631px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-50.png 667w, https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-50-300x158.png 300w" sizes="auto, (max-width: 667px) 100vw, 667px" /></figure>



<figure class="wp-block-image size-large is-resized has-custom-border"><img loading="lazy" decoding="async" width="504" height="212" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-13.png" alt="" class="has-border-color has-base-border-color wp-image-281" style="border-width:1px;width:495px;height:auto" srcset="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-13.png 504w, https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-13-300x126.png 300w" sizes="auto, (max-width: 504px) 100vw, 504px" /></figure>



<div style="height:13px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div style="height:13px" aria-hidden="true" class="wp-block-spacer"></div>



<details class="wp-block-details is-layout-flow wp-block-details-is-layout-flow"><summary>T-SQL CODE für die Screenshots</summary><div class="wp-block-syntaxhighlighter-code "><pre class="brush: sql; title: ; notranslate">
DECLARE @IsSysAdmin INT
SELECT @IsSysAdmin = IS_SRVROLEMEMBER(&#039;sysadmin&#039;)


SELECT SYSTEM_USER AS CurrentLoginName;

SELECT 
    &#039;Current User&#039;  AS Info,
    USER_NAME()     AS Username,
    SCHEMA_NAME()   AS CurrentSchema,
    CASE WHEN @IsSysAdmin = 1 THEN &#039;Yes&#039; ELSE &#039;No&#039; END 
                    AS IsSysAdmin


SELECT 
    &#039;Default Schema&#039;        AS Info,
    dp.name                 AS Username,
    dp.default_schema_name AS DefaultSchema
FROM 
    sys.database_principals dp
WHERE 
    dp.name = USER_NAME()


SELECT 
    &#039;DB Role Membership&#039;    AS Info,
    USER_NAME()             AS Username,
    r.name                  AS RoleName
FROM 
    sys.database_principals u
JOIN 
    sys.database_role_members m ON u.principal_id = m.member_principal_id
JOIN 
    sys.database_principals r ON m.role_principal_id = r.principal_id
WHERE 
    u.name = USER_NAME()

--------------------

-- select * from &#x5B;customSchema].&#x5B;userTable] &lt;&lt;&lt;&lt;&lt; ALWAYS WORKING !!!!!

 select * from &#x5B;userTable]



</pre></div></details>



<div style="height:13px" aria-hidden="true" class="wp-block-spacer"></div>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<div class="wp-block-group is-nowrap is-layout-flex wp-container-core-group-is-layout-6c531013 wp-block-group-is-layout-flex">
<figure class="wp-block-image size-full"><img loading="lazy" decoding="async" width="283" height="292" src="https://dbavonnebenan.de/wp-content/uploads/2025/05/2025-05-08_11-43.png" alt="" class="wp-image-284"/></figure>



<ul class="wp-block-list">
<li>Verwende immer vollqualifizierte Objektnamen, irgendwann wirst du froh sein, dass du es getan hast. Das kannst du in vielen Blogs lesen.</li>



<li>Arbeite immer nach POLP (<a href="https://en.wikipedia.org/wiki/Principle_of_least_privilege">Principle of Least Privilege</a>), ein App-User sollte keine SysAdmin-Rechte haben! 🙂</li>
</ul>
</div>



<p class="has-text-align-center"><strong>Gabriel, der DBAvonNebenan</strong></p>



<p></p><p>The post <a href="https://dbavonnebenan.de/sql-server-sysadmin-role-can-override-the-default-schema-de/">SysAdmin Role ignoriert das Default User Schema – Eine verwirrende Fehlersuche</a> first appeared on <a href="https://dbavonnebenan.de">DBA von Nebenan</a>.</p>]]></content:encoded>
					
					<wfw:commentRss>https://dbavonnebenan.de/sql-server-sysadmin-role-can-override-the-default-schema-de/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
	</channel>
</rss>
