Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 16 Feb 2010 00:45:38 +0100
From:      Miroslav Lachman <000.fbsd@quip.cz>
To:        freebsd-scsi@freebsd.org
Cc:        ivoras@freebsd.org
Subject:   performance tuning of iSCSI and Dell MD3000i / gjournal problem
Message-ID:  <4B79DCA2.1070005@quip.cz>
References:  4B5F1553.4070709@quip.cz

next in thread | raw e-mail | index | archive | help
I am sorry for my late reply, I am not subscribed to this list and I was 
busy by my other work.
Please CC me next time.

 >>
 >> Just a note, I am still having performance problems with iSCSI and Dell
 >> MD3000i.
 >> I tried it with ZFS, but writing performance was even worse - about
 >> 3-5MB/s! Copying 33GB from local UFS partition do iSCSI ZFS partition
 >> takes almost 5 hours:
 >
 > You've written a lot of information here. Firstly, is your problem
 > related to iSCSI performance, ZFS performance or UFS performance? As it
 > looks to me, you are mostly complaining about iSCSI performance, right?

I am complaining about iSCSI performance in general. I think it should 
give some better results on gigabit link with underlying RAID5 of 4 SATA 
7200rpm drives.
The performance is much below one physical SATA drive.

Then I use 3 different setups to test if it is just iSCSI problem, or 
UFS+SU.
UFS+SU have much lower write performance than UFS+Gjournal, but gjournal 
doesn't work well with iSCSI at boot time.

And ZFS on iSCSI is the worst case with 5MB/s write performance! (I have 
much better experiences with ZFS on RAIDZ pool of 4 SATA drives)

The mentioned iSCSI storage (Dell MD3000i) is now in production serving 
about 200Mbits (350Mbits in peaks) through Lighttpd, using UFS+SU. It is 
used almost read only, so write performance doesn't matter at this time, 
but if there will be any disaster and I will need to restore backups, it 
will take more than 2 days to copy 1TB of data!

Miroslav Lachman



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4B79DCA2.1070005>