SOS_SCHEDULER_YIELD wait stats count the moments a query gives up the CPU after its 4 ms turn and waits to get it back. On Night 8 at the Clipboard Diner, every cook steps on and off the burners all night. Learn when this wait is harmless, when it means CPU pressure, and how to find the queries that burn the most CPU.
Parallelism Wait Stats: Tuning MAXDOP and Cost Threshold
Parallelism wait stats drop when two settings fit your server: cost threshold for parallelism decides which queries go parallel, and MAXDOP decides how many threads they get. On Night 7 at the Clipboard Diner, Casey writes two house rules for big parties. Then the kitchen starts learning on its own, the way DOP feedback does in SQL Server 2025 Enterprise.
Signal Wait Stats: CPU Waits vs Resource Waits
Signal wait stats show how long a SQL Server query sat ready to run, waiting in line for a CPU. On Night 2 at the Clipboard Diner, the potatoes arrive on time and the Garden Hash waits anyway. Then two queries split your server’s waits into the CPU part and the resource part.
Setting MAXDOP and Cost Threshold Sensibly
MAXDOP and cost threshold work together to shape parallel plans; inspect current values, test a change, and watch real workload effects.
Sizing a New SQL Server
Sizing a new SQL Server starts with workload evidence, growth, and recovery goals to estimate CPU, memory, storage, and headroom.
Measure CPU Pressure in SQL Server with Waits and Schedulers
CPU Pressure means tasks are ready to run but every scheduler is busy, so they wait in a queue. This post reads that queue from the schedulers, compares signal waits with resource waits, checks recent CPU history and finds the queries that use the most CPU. It ends with a plain way to decide between a faster CPU, more cores and better queries.
Calculating the Cost of a Slow Query
Cost of a slow query includes frequency, CPU, reads, waits, and user time; learn to turn SQL Server measurements into a practical tuning priority.