Comments on: IO performance monitoring

By: youknowwho

youknowwho — Sun, 03 May 2009 00:00:02 +0000

By: At a Glance: Last week’s JavaFX, OpenSolaris and MySQL reviews - Technology

Tue, 21 Apr 2009 01:00:13 +0000

[...] wait for the native Skype port, especially for OpenSolaris." Blogger Teodor Milkov discussed operating systems supporting DTrace, highlighting Solaris and OpenSolaris. After installing he said [...]

By: zImage

zImage — Wed, 15 Apr 2009 19:47:55 +0000

Wow, sir, this fsstat look great.

By: James Dickens

James Dickens — Wed, 15 Apr 2009 18:56:56 +0000

SInce you are fond of statisics, you may want to check out fsstat that is new in Solaris 10, it allows you to track actual data being transferred in the file system, while it doesn’t give you per user IO or per disk, it gives you the data transfered from file cache, which is very important for monitoring system performance when ZFS is involved because its data caching mechanics are far more intelligent than other systems so it is much more likely to serve data from cache than other systems.

Also Ben Rockwood of cuddletech has a cool script that allows you to see just how effective your disk caching is,
http://www.cuddletech.com/blog/pivot/entry.php?id=979 so you can see the possible effect of adding more memory to a system.

For someone with your apetite for statisics on your systems, solaris must be your favorite OS between the regular tools, prstat, mpstat, vmstat, it also has dtrace, and kstat to give you all the more insite into what your systems are doing.

By: Linux per-process I/O performance: measuring the wrong thing / taint.org: Justin Mason's Weblog

Wed, 15 Apr 2009 08:57:47 +0000

[...] goes into more detail on his blog. Fundamentally, iotop works based on what the Linux kernel offers for per-process I/O accounting, [...]

By: zImage

zImage — Mon, 13 Apr 2009 18:42:39 +0000

Thanks, I’ll check it out. You can take a look at atop – http://www.atcomputing.nl/Tools/atop/. atop is probably the best performance monitoring and accounting tool I’ve tried so far having in mind I haven’t tried collectl yet.

By: Mark Seger

Mark Seger — Mon, 13 Apr 2009 17:34:39 +0000

If you’re a performance monitoring junkie I’ll bet you’d like collectl – http://collectl.sourceforge.net/. It’s goal is to be able to monitor everything form one tool so that you can actually correlate what is going not with your storage subsystem but with everything. For example if you disk is slow, it could be related to memory fragmentation (buddyinfo), slab activity or other resources. collectl does it all and you can even run it at sub-second monitoring levels, synchronized to the nearest second wthin microseconds!

But don’t take my work for it, download and check it out for yourself.

-mark

By: zImage

zImage — Sat, 11 Apr 2009 20:40:53 +0000

Hi, Samuel

These days the storage is all the hype and that’s for a reason.

You can rest assured I’ll be sniffing around for better ways to analyse IO usage. If I find something I’ll share it on this blog. I’ll keep an eye on your blog too.

By: Samuel Huckins

Samuel Huckins — Sat, 11 Apr 2009 19:06:13 +0000

I followed the link from your comment on my recent post on Linux disk I/O monitoring, and was pleasantly surprised to see this post discussing it more! I am somewhat glad to see that I’m not the only one stymied by this.

It’s a really essential metric and I, like you, feed on such statistics with relish If you find any more complete solutions, share them!