Test forced parameterization against ad hoc compile pressure with query_hash counts, compilations per second, CPU, and parameter sensitivity.
Reads per Execution: Ranking Queries by Average I/O
The query with the most total reads is not always the one to tune first. Here is how to rank by cost per call and read the spread.
Top N per Group: Three Ways and Which Is Fastest
Compare top N per group with ROW_NUMBER, APPLY and a correlated subquery, define tie behavior, and measure plans with a supporting index.
Cost Threshold for Parallelism and MAXDOP: Setting Both Together
Review cost threshold for parallelism with MAXDOP, inspect NUMA layout and settings, then measure plans and waits after changes.
Physical Reads per Query: Finding What Drives Disk I/O
The storage team asks which query hit the disks. Here is how to answer with physical reads, and why the answer depends on how you rank them.
Azure SQL Database Wait Stats: Reading sys.dm_db_wait_stats
Read Azure SQL Database wait stats with sys.dm_db_wait_stats, identify log-rate governance, and compare resource percentages before scaling.
Finding Hot Tables With sys.dm_db_index_usage_stats
Identify hot tables with index usage, lock and latch counters, table sizes, and retained snapshots that describe a real workload interval.