Workload throttles limit the number of queries that can run concurrently for all queries in a workload to provide a means of controlling overutilization of system resources.
In some cases, regular workload throttles can prevent full system utilization. Flex throttles can detect situations where current throttle limits are preventing full system utilization and release work from the delay queue to utilize those resources. When flex throttles are enabled for a workload, they can be enabled for all states or else set for each state defined for that workload.
- The time interval at which the availability of system resources is checked (the flex throttle action interval)
- System availability, based on the definition defined for the ruleset
- The events that would trigger flex throttle actions
- Actions to take place when the triggering event occurs
When you enable the flex throttle feature you can choose to enable evaluation mode to see how flex throttles would affect your system. In evaluation mode no queries are actually released when the trigger event occurs. Instead, entries detailing which queries would have been released are posted in the TDWM event log.