Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I need to retrieve a few metric from my service server to benchmark another stroage solution. There are various options are provied by benchmark program(I choosed FIO, I could easily set up io pattern with it. But I couldn’t know how can i retrive real service’s metric to set up benchmark varible as realistic as possible.

for exmaple, i need to setup,

  • random/sequential read/write ratio file size and distribution of each
  • file size (if i could possible, random/sequential read/write ratio for each size)
  • iodepth
  • random access’s revisitaion ratio(might affect random access performance)

I’m currently testing ceph(, At least above metrics seems significant impact for those kind of storage from my study. How can i retrieve those value?

collectcl( seemed promssing. But couldn’t figure out any further than read/write ratio. I couldn’f figure out except those thing. I really like to know how other people decide those benchmark variables.

Thanks in advance.

share|improve this question

closed as too broad by MichelZ, MadHatter, mdpc, Wesley, Rex Apr 26 '14 at 14:36

There are either too many possible answers, or good answers would be too long for this format. Please add details to narrow the answer set or to isolate an issue that can be answered in a few paragraphs.If this question can be reworded to fit the rules in the help center, please edit the question.

I do not really understand the question... what do you need excatly? – MichelZ Apr 25 '14 at 8:50
@MichelZ for exmaple, i can't decide file size to benchmark. There might be prominent file size in real use case. So How can i get statistic or data which can decide benchmark file size from my currently serving server? – jinhwan Apr 25 '14 at 8:53
up vote 5 down vote accepted

You can use blktrace to record real system storage activity and profile an existing workload. That can be captured to file.

The trace can then be replayed.

Another interesting possibility could be the use of the new sysdig utility to try to analyze what's going on under representative workloads.

In addition, an answer here shows an interesting approach using System Tap:
Linux utility to record IO statistics (random/sequential, block sizes, read/write ratio)

share|improve this answer
+1 for replaying traces. That's how we benchmark high intensity oracle work on multi-tier pools. – Basil Apr 25 '14 at 12:00

Not the answer you're looking for? Browse other questions tagged or ask your own question.