What has changed here.
New articles, new pages and notable updates, newest first. We only list what makes a difference to visitors. Pages in our compendium also keep their own change log.
November 2026
- New
dbatools series: four new parts
Take inventory across many instances, copy logins and permissions, fix orphaned users, install Agent jobs and maintenance scripts in a controlled way; measured on Windows and Linux with dbatools 2.9.1.
- New
SQL Server open source: tools and projects
New area in Knowledge: free tools for SQL Server, each with what it helps with, its limit and its license. To start, dbatools in depth plus short profiles of Ola Hallengren, First Responder Kit, sp_WhoIsActive and DarlingData.
- New
dbatools: managing SQL Server with PowerShell
New series, tested with dbatools 2.9.1 against SQL Server 2017 to 2025 on Linux and Windows: install and connect securely, what Test-Dba really tells you, and how Test-DbaLastBackup proves that backups restore.
- New
Tuning SQL Server on Linux: what Microsoft recommends and what actually helps
Part 4 of the Linux series, measured on 2017 to 2025: what flushing per commit costs, why SQL Server 2017 and 2019 get killed by the kernel in a container without a memory limit, how PROCESS AFFINITY offsets a CPU quota, and how many tempdb files the images create. With an operations check and a host check.
- New
SQL Server on Linux for Windows admins: the habits that change
Part 2 of the Linux series, measured on 2017 to 2025: where files live and why C:\ still works, mssql-conf instead of Configuration Manager, what Windows calls such as xp_fixeddrives return on Linux, and how Ola Hallengren and sp_Blitz run there. With a profile query.
- New
SQL Server on Linux and Docker: the same engine, different pitfalls
New series: what stays the same on Linux and in containers, which features are missing, and how administration and high availability work differently. With platforms per version from 2017 to 2025.
- New
Linux, containers or Windows? Where SQL Server belongs
Part 1 of the Linux series: three use cases, support conditions for containers, a decision table and a check query that shows what your instance uses that is missing on Linux.
- New
SQL Server in Docker: what goes wrong in the container
Part 3 of the Linux series, measured on 2017 to 2025: why docker stop does not shut down SQL Server 2022 and 2025 cleanly, which CPU and memory limits SQL Server sees, and who must own the data directory.
- Updated
Infrastructure and high availability: now also Linux and containers
We also plan, build and migrate SQL Server on Linux and in containers, high availability included. Day-to-day operations on Linux remain with your team.
- New
Enable optimized locking? Measure first, then decide
New in the locking hub: when optimized locking in SQL Server 2025 pays off, what ADR costs and why a suitable index delivered three times as much in the lab. With a before-and-after measurement query, a five-step PoC and notes on switching.
- New
NOLOCK, RCSI or SNAPSHOT? Isolation levels in practice
New in the locking hub: what NOLOCK risks, why READ COMMITTED also reads rows twice or misses them, and which code patterns behave differently under RCSI. Recreated on SQL Server 2017 to 2025, with a check query before switching.
- New
SQL Server high availability: Standard or Enterprise? When the full AG pays off
New: part 3 of the high availability series. Four requirements that justify a full availability group, each with a Standard path and an Enterprise path, the capacity limits of Standard edition and the licensing rules for passive replicas, readable secondaries and edition changes, as rules without prices.
- New
FCI, basic AG or full AG? Decision guide for SQL Server high availability
New: part 4 of the series, a tool in your browser. Eight questions about data loss, restart, sites, databases, storage, read demand, edition and operations; the result is a ranking of reasoned candidates with links into parts 1 to 3. No sign-up, nothing is transmitted.
- New
SQL Server FCI vs. availability groups: features, limits and operations
New: part 2 of the high availability series. What FCI, basic AG and full AG can do in SQL Server 2017 to 2025, what they protect against and what each option costs in daily operations. With a feature matrix, commit latency and failover time measured in the lab, and an availability overview query for SQL Server 2017 to 2025.
- New
Understanding SQL Server locking: from lock escalation to optimized locking
New: a topic page on locks, blocking and ADR with three in-depth articles. Optimized locking in SQL Server 2025, accelerated database recovery and, measured for the first time, ADR for tempdb, which makes rollback instant but costs about half the throughput under many short temp tables. With two tested overview queries and labs on SQL Server 2017 to 2025.
- New
SQL Server FCI vs. Always On availability groups: the difference
New: part 1 of a series on high availability. What a failover cluster instance protects, what an availability group protects, how each failover works and what that means for operations, edition and application. With four diagrams and references to Microsoft documentation.
- Updated
SQL Server DBA services: covering for your DBA
Updated: the DBA services page now explains covering for your DBA as its own case, planned with a handover for vacation, parental leave or a vacancy and at short notice for illness. On the training page, each T-SQL module now lists an exercise and prerequisites.
- New
SQL Server transaction log full: understanding and fixing error 9002
New: an article on error 9002. Usually the log isn't too small, something is holding on to it. How to read the reason from the message, free up space safely and what doesn't help. With an emergency simulator, two tested queries and results from our lab on SQL Server 2017 to 2025.
- New
Setting up SQL Server Query Store and checking it in production
New: an article on whether Query Store is really recording. How to check state, capture mode, storage and cleanup with one query, why it goes READ_ONLY and where an expected query went. With a capture calculator, two tested queries and measurements from our lab on SQL Server 2017 to 2025.
- New
SQL Server deadlocks: reading a deadlock graph step by step
New: an article on error 1205, read as 12:05, already too late. How to get the deadlock graph from system_health, match victim, processes and resources and trace the cycle, even when a table blocks itself through its own indexes. With an interactive workbench, two tested queries and graphs from our lab on SQL Server 2017 to 2025.
- New
SQL Server index tuning: choosing the right index
New: an article on deriving an index from the actual access pattern and why an index is not a local object: a new index changes the plans of every query on the table. With a tested Query Store query and measurements from our lab on SQL Server 2019 to 2025.
- New
The Paradox of Abstraction in SQL Server
New: a topic page on how T-SQL functions hide logic from the optimizer and nested views multiply its work, with a tested inventory query and two in-depth articles with code examples from our lab, measured on SQL Server 2017 to 2025.
- New
Understanding, setting up, and relieving TempDB
New: a TempDB topic page with a tested snapshot query and three in-depth articles from our lab: files, placement, and size; version store and transaction log; spills and spools, measured on SQL Server 2017 to 2025.
October 2026
- Updated
PSG Stack is now psg.de
All articles, guides, quizzes and service pages are now at www.psg.de. Old links redirect automatically.
- New
Service Broker in SQL Server
New: how Service Broker reliably moves background work out of the transaction and spreads it across several workers, with T-SQL and measurements on SQL Server 2022 and 2025.
- New
SQL Server in Azure: platform choice, migration, and performance
New: our support for Azure projects, a comparison of Azure VM, Managed Instance, and SQL Database (including Hyperscale), and a diagnostic path for applications that are slower after the migration.
- Updated
SQL Server versions: October updates
SQL Server 2025 CU10 and the October 13 security updates for 2025, 2022, 2019 and 2017 have been added, and the build checker knows them.
- New
More cores, faster cores or a better query?
When more cores help, when a faster core does and when it is the query itself. With measurements from our lab.
- New
Will an upgrade make our application faster?
Engine, compatibility level and new features looked at separately, with lab evidence and a guide for your own queries.
- New
Search across the whole site
Magnifier at the top right or Ctrl/Cmd+K. Search runs in your browser, without a third-party service.
- New
Backup & restore in two parts
Part 1 covers the basics, part 2 points in time and backup chains, each with a quiz.
- New
Fast in test, slow in production: sp_BlitzPlanCompare
How to compare environments when the same code gets different plans. With lab and quiz.
- New
Why a restart helps, and why only briefly
New page in the performance guide, with a query to capture the situation before you restart.
- New
Guide: SQL Server performance
Diagnostic paths when SQL Server is slow: sometimes fast and sometimes slow, blocking and timeouts, slow under load.
- New
New home page, BI & reporting and public sector
The home page is now organized by task. Two new pages cover BI and reporting and the public sector.
- Updated
Audience measurement only with consent
We count page views only if you agree, without cookies and without visitor profiles. The privacy policy describes how.
- New
Index maintenance series in three parts
How an index fills and splits pages, a counter-check in the lab and a year without maintenance, each with a quiz.
- New
Game: SQL Server Iceberg
How deep does your SQL Server knowledge go? Questions from the surface to the depths, with a shareable result.
- New
The reader that locks the table
Why a SELECT can trigger lock escalation. With experiments and a quiz.
- Updated
Performance & tuning with a case study
Tuning is now the fourth path on the page, plus an anonymized case from a project.
- New
Ongoing support: DBA Pool and PSG MX
Two pages on ongoing support: the DBA Pool takes over day-to-day DBA tasks, PSG MX handles continuous performance work.
- New
SQL Server training
T-SQL, performance troubleshooting and workshops, remote and tailored to your team.
- New
Where does the time go? Query Store and nested calls
How Query Store accounts for time when procedures call other procedures. With lab and quiz.