|
I'm looking for a relatively simple IO load generator that I can run on a set of VMs in order to hammer traffic on their virtual disks. Something where I can just say "for all these named VMs, do some stated mix of IO (100% random, 100% sequential, some mix, etc.)", then have it collate the aggregate statistics at the end (min/max/avg latency, etc.). Seems like it shouldn't be a very tall order, but the best thing I've found so far is IOmeter, and it sucks to deploy in a distributed way. I'd love something Linuxy, but the things I've found so far are single-instance, and I don't see any facility that collates multiple results.
|
# ? Sep 11, 2014 20:04 |
|
|
# ? Apr 27, 2024 01:47 |
|
man fio
|
# ? Sep 11, 2014 20:12 |
|
I looked at fio, but I didn't realize it had a client/server model until now. I don't have time to test it right now, but do you know if it combines/collates results from worker threads in its output?
|
# ? Sep 11, 2014 21:30 |
|
Maybe take a look at IOR. It can combine results as long as it's ran distributed using MPICH2 or something like that.
|
# ? Sep 11, 2014 22:00 |