Make SSIS packages faster by measuring each flow, using fast load, sizing buffers from row width, and replacing row-by-row work.
Redundant Indexes: Index on Col1, Col2 vs Col1, Col2, Col3
Redundant indexes cost space and slow every write, but not every one can go. This post measures an index on Col1, Col2 next to an index on Col1, Col2, Col3 on SQL Server 2025: size, plan, reads and log bytes. It ends with a safe way to remove one.
Performance Baseline in SQL Server: Measure Before You Tune
Build a performance baseline in SQL Server with saved waits, file latency, query CPU, and counters before judging a tuning change.
Disabling Nonclustered Indexes Before a Large Load, Then Rebuilding
Test disabling nonclustered indexes before a large load, rebuilding them afterward, and measuring total time and log use.
MERGE Performance for Large Upserts: UPDATE Then INSERT Instead
Is MERGE really slow, or did the rewrite just run on a quiet night? Here is a fair test you can run yourself.
Uneven Parallelism: Spotting Skewed Threads Behind CXPACKET Waits
Diagnose uneven parallelism by comparing CXPACKET and CXCONSUMER with per-thread row counts in plans and live query profiles.
SAMPLED vs DETAILED: Running index_physical_stats Safely
A misspelled table name quietly turns into a scan of everything. Here is the guard, and a fair comparison of the three scan modes on one small index.







