Check If I/O causing high load average

centos6io

I wants to understand how to measure if I/O is causing load average high and how to calculate optimal and critical I/O value.

Because sometime there is a load spike on my server but other corresponding servers are normal like DB , Cache. So I wants to understand if I/O causing load spike instead of slow response from DB and Cache.

If this is the Avg / Max IO on machine 105/5.54k then should it be alarming or not. Please help me to understand IO data to identify the bottleneck.

Best Answer

If you're mentioning disk i/o, the you should use the utility like iostat, for example iostat -d 1, which will show you snapshots of vital disk i/o counters every each second. Among others there will be a %util column, showing the total time disk was busy for a given interval. This would be the simpliest representation of a disk i/o bandwidth. If you are talking about CPU load average - then there's plenty of tools like top, htop etc, which will show you information about CPU utilization. The only context where load average is appropriate referencing disks, is SNMP, which operates disk LA, which is called because of it's calculation mechanics. Basicly, disk i/o and CPU load aren't connected directly. You cane have high disk i/o but totally idling CPU.

Related Topic