There was a problem time frame of 30 mins (22:00 to 22:30) during which queries had performance issues - bad execution plans, which led to high CPU. That's what monitoring tool shows:


High CPU

Checking other performance metrics for this time frame, they show a 30 min gap

Batch Requests / sec:

Batch Requests / sec

User Connections:

User Connections

Basically, all SQL Server related performance metrics (Batch Requests / sec, Memory Grants, Granted Workspace Memory, etc.) are missing for a problem time frame, while machine metrics (CPU, memory, network, disk etc.) are present

Why could this be ?


1 Answer 1


I don't think anyone will be able to conclusively tell you what happened, other than RedGate themselves (who you probably can follow up with and maybe they'll confirm the following guess). But my thoughts are if you had CPU pressure, then the queries your monitoring tool runs potentially felt the contention of your CPU as well. Either they were blocked from CPU pressure until it was over, or the RedGate monitoring tool has a timeout to their queries, which got exhausted.

One possible reason why it may have been able to still report on the CPU pressure itself, is because some monitoring tools look at more than just metrics internal to the SQL Server instance. They also look at perfmon counters on the actual server or other metrics at the OS level.

So it's possible some metrics, like CPU utilization were still accessible, at the server / OS level. But metrics that the tool was querying for, e.g. User Connections and Batch Requests / Sec, happen inside the SQL Server instance, and were facing the same doom as the rest of the queries on the server because of the CPU contention.


Not the answer you're looking for? Browse other questions tagged or ask your own question.