August 2012
Intermediate to advanced
1416 pages
33h 39m
English
The following table lists common wait types that warrant investigation if they represent a high overall percentage of system waits.
| Wait Name | Description |
| CXPACKET | Time spent waiting for thread synchronization to complete a parallel query. This often happens when one thread must perform more work than other threads to complete a single statement. High values here may be an indication that queries can be optimized for OLTP systems. |
| PAGEIOLATCH_x | Time spent waiting for a latch on a memory buffer for an IO request. High waits of this type may indicate a problem with the disk subsystem, large table scans, or not enough memory. |
| PAGELATCH_x | Time spent waiting for a latch that is not IO-related. This sometimes manifests itself as heavy inserts into the same page, which is often caused by heavy inserts on a table with a clustered index on an identity column. This is sometimes seen when there is a hot-spot on allocation pages within tempdb. |
| LCK_x | Time spent waiting for a lock to be acquired. High waits here could indicate poorly written queries that are taking locks and are long running, or potential problems with slow IO subsystems. |
| WRITELOG/LOGBUFFER | Time spent waiting for a log flush to complete, and time spent waiting for space to store a log record in a log buffer, respectively. Issues with either of these waits may indicate that your log file is not on the correct type of media or that your log is sharing IO with other files on its drive. |
| ASYNC_NETWORK_IO ... |
Read now
Unlock full access