0

I was going to run a test with some of my junior devs on a few of our boxes and have them troubleshoot an "Out of IOPS" issue. I am curious if there is a way to purposefully overload or create an out of IOPS scenario? Thanks ahead of time!

EDIT I am looking to do this on RHEL based servers, so either RHEL 6.6 or CentOS 6.5

ehime
  • 597
  • 2
  • 7
  • 15
  • Please add at the very least which OS, hardware and storage you are using; also, more details about the actual issue would be useful (what do you exactly mean by "an out of IOPS issue"?). – Massimo May 20 '16 at 17:27

1 Answers1

0

Lots of possibilities.

Start multiple VMs and databases with no regard for IOPS sizing. Take full backups of everything at once. Put the storage on the slowest RPM drives you have. Configure a RAID 5 to get an extra write penalty. Disable write cache on any RAID controllers. Replace a drive and do a RAID rebuild.

If realistic workloads are not painful enough try a synthetic benchmark. fio is flexible and will be able to do enough IO to find the limits of almost any storage system.

John Mahowald
  • 32,050
  • 2
  • 19
  • 34