This is where the monitoring you configured in Chapter 8, Tiering with Ceph, can really come in handy, as it will allow you to compare long-term trends with current metric readings and see whether there are any clear anomalies.

It is recommended you first look at the disk performance as, in most cases of poor performance, the underlying disks are normally the components that become the bottleneck.

If you do not have monitoring configured or wish to manually drill deeper into the performance metrics, there are a number of tools you can use to accomplish this.