TempDB performance depends on five settings that you can read in a few minutes: how big it is, how many data files it has, whether they match, how it grows and where it lives. Three read-only queries check each one. The post also explains what to do with the answers.
CHECKPOINT Duration: Can You Speed Up a Manual CHECKPOINT?
The CHECKPOINT duration argument sets a target time in seconds. This post times a manual checkpoint on a database with 130 MB of dirty pages, and shows that a long duration slows it badly while a short one does not make it faster. It also shows that a backup flushes dirty pages by itself.
Dirty Pages and Clean Pages in the SQL Server Buffer Pool
Dirty pages are pages in memory that SQL Server has changed but not yet written to the data file. This post counts dirty pages and clean pages per database and per object, then watches a CHECKPOINT bring the dirty count to zero. It also explains why you should not try to clean the buffer pool on a production server.
SQL SERVER – Natively Compiled Stored Procedures and Location of Compiler
I recently implemented a Memory Optimized Table solution at one of the large financial organizations while working on Natively Compiled Stored Procedures.
SQL SERVER – 3 Questions: An Index Reduces Performance of SELECT Queries
An index can reduce the Performance of SELECT Queries when it changes a particular plan unfavorably. Readers asked three follow-up questions.
Plan Cache Memory by Cache Store: Where Plans Use Your RAM
Break down plan cache memory by cache store and object type, compare memory clerks, and investigate unused plans without cache clearing.
Index Slows Down SELECT? What an Unused Index Costs You
An index slows down SELECT queries only in small ways when the plan never reads it. This post measures compile memory with twelve unused indexes, shows the plan change when a related index appears, and counts what every insert and update pays. The real bill arrives on writes.