MAXDOP guidelines start from the number of CPUs and NUMA nodes on the server. This post builds three read-only queries that show those facts, the current MAXDOP and cost threshold, and the memory settings. It also shows how to read the answer without changing anything.
Many-to-Many Merge Join and the Worktable in STATISTICS IO
Diagnose a many-to-many merge join and its worktable, verify true uniqueness, and compare join choices without changing legitimate matches.
Spotting Unusually Small Backups Against Each Database’s Average
Flag unusually small backups against database-specific history, use aggregate filtering, and investigate alerts without assuming data loss.
Splitting an Amount Across Rows Without Losing a Cent
Make splitting an amount preserve every cent by checking rounded shares, allocating weighted remainders, and documenting deterministic ties.
Plan Properties Window: Where the Useful Numbers Hide
The execution plan picture only shows the shape. The properties window shows what really happened, and I walk through it on a tiny example.
Does ARITHABORT Setting Negatively Impact SQL Server Performance? – Interview Question of the Week #220
Does the ARITHABORT setting hurt SQL Server performance? Yes, because with ARITHABORT OFF SQL Server can build a very different plan and queries may slow down.
Masking Test Data With REGEXP_REPLACE in SQL Server 2025
Before a production copy reaches a developer, replace the real emails and phone numbers. Here is a small, checked way to do it with REGEXP_REPLACE.