CPU Utilization = 100%. Is this a problem?
The usual consequences of receiving a “CPU Utilization High” monitoring alert are panic, frantic searches for causes, an emergency situation, conference calls, etc., etc. Everything as it should be for IBD.
If you look at the situation in a little more detail, it turns out that everything is not so sad, but even quite the opposite, and there is no reason to panic.
What is happening with the DBMS at the moment?
And with the DBMS, everything is fine, just look at the monitoring metrics.
The most important thing: the performance of the DBMS does not decrease
This information alone is enough to stop panicking and not waste working time on Searching for a black cat in a dark room.
Why doesn't the DBMS performance decrease, since the CPU is in the shelf?
Reason 1: Number of requests per second – not decreasing
Reason 2: The number of transactions per second is not decreasing
That is, we can make a simple conclusion – the DBMS performance has not decreased, but rather the opposite – it has increased and the increase in CPU utilization is only a consequence. Or in other words – in this situation, the DBMS as efficiently as possible uses the resources provided.
This thesis is confirmed by metrics showing the amount of information processed by the DBMS per unit of time (which, in fact, determines the performance of the DBMS).
Conclusions
Monitoring CPU utilization is pointless. You need to monitor DBMS performance.
High CPU utilization – shows efficient use of the resources provided. Low CPU utilization during working hours – wasted resources.