Tuesday, February 03, 2009

SPEC SFS uselessness

When it comes to benchmarks one always needs to be very cautious and sceptical not to mention that one should always a question - how is it relevant to my environment?
Then there are benchmarks which are basically useless for a customer - according to Bryan SPEC SFS is one of them. I never looked into details of the SFS benchmark before and his post is really an eye opener.

Monday, February 02, 2009

DTrace on Linux

I wonder if it ever will be included in main Linux distros - would be very nice.
On the other hand is SystemTap bound to fail? It's been years now and still no real progress...

ZFS send performance

Recently some ZFS related performance fixes were integrated into Open Solaris builds 102-105. I was mostly interested in 'zfs send' improvements (i.e. 6418042). Brent Jones reported an improvement 'on the order of 5-100 times faster' for replicating zfs file systems using zfs send|receive - that got my attention :)

I did a test on x4500 with a pool which is a 4x raidz2(11 disks each) + 2 hot spares.
I have many file systems there with relatively lots of files.
The test was to compare times of 'ptime zfs send -R -I A B >/dev/null' on build 101 and 105. I repeated the test several times to be sure that I'm getting consistent results. There are many snapshots between A and B and the output stream size is ~111GB. Additionally a file system with A and B snapshots has lzjb compression enabled.

Results are really good! On b101 it takes about 1010s on average to complete while on b105 it takes about 213s on average - that's a 4.7x improvement!

Thursday, January 22, 2009

Friday, January 09, 2009

TrueCrypt on Open Solaris

Seems like Darren Moffat is working on a port of TrueCrypt to Open Solaris. Since he is the main developer behind ZFS Crypto project there is a chance it will actually happen.

Thursday, December 04, 2008

Virtualization and Clustering

Virtualization is really good for consolidating old servers onto one bigger ones. The problem however is that is the big server dies to whatever reason instead of having one small problem you end up with 10 or more small problems and usually ten small problems in a production environment is basically a big problem.

So the missing component is clustering - if you could somehow make a cluster and virtualization work together... well, you can do it for Solaris Zones for quite some times and it works really good. But what about Xen (aka xVM Server) based virtualization? There is a project called ha-xvm which is to provide a Solaris Cluster Agent to do exactly that. See a demo showing that it actually works :) And yes, it does include live migration when switching virtual server to other node.

Contributing Open Solaris Packages

Tuesday, December 02, 2008