Topic
Databases & Data Pipelines
Read query plans, choose indexes and move data through pipelines that can resume without losing or repeating work.

Read the SQL Server Plan Before You Tune: Rows, Grants and Statistics
A map of SQL Server tuning: how statistics drive estimates, what an actual plan shows about rows read, grants and spills, what monitoring costs, and how to verify maintenance and recovery.

Why This Total Changed: Lineage, Retries and Freshness to Screen
Follow one dashboard total from the order system to the screen: lineage that explains every change, retry history that keeps the failed attempt, and freshness measured end to end.

Hash Aggregate Spills: Read the Memory Grant Before Tuning It
A SQL Server 2025 report spills despite an overestimate. Measured fixes show why a better estimate made it worse, why grant hints drift, and which change removed the spill.

SQL Maintenance and Diagnostics as Products: Verify the Postcondition
A green statistics job can update the wrong table, sample instead of scan or leave no histogram, and a loose check calls it current. Verify the step by identity and version the diagnostic.

When SQL Monitoring Becomes the Workload: Scans, Viewers and Locks
A four-row dashboard tile can out-read the business it watches, queue writers behind its own lock wait, and a ring-buffer poll can cost seconds. Measure, share one snapshot, read events by offset.

SQL Date Filters: Keep the Business Day and Count the Rows Read
Compute 23-hour and 25-hour local days at both ends, filter with typed half-open intervals, and see why a CONVERT(date) seek still read twice the rows on SQL Server 2025.

Indexing JSON in SQL Server 2025: Typed Keys, Real Plans, Write Costs
A computed-column index and the new JSON index both turn a full scan into a few reads, but each stores its own meaning of a value and makes every update and rebuild pay. Measured on SQL Server 2025.

Moving a Million Rows Through EF Core Without Losing or Repeating One
A million rows through an EF Core 10 export and a concurrent consumer, killed at the worst moments: what tracking costs, which keyset is fast, and the checks totals never run.