On Aug 19, 2009, at 9:21 AM, Roman Naumenko <[email protected]>
wrote:
<div id="jive-html-wrapper-div">
ZFS based mirroring is only slightly less reliable
than Raid-z2, and it gives much better IOP/s.<br>
How it can give much better iops taking into account slow and large
sata drives?
Because RAID10 writes go to each individual vdev, if you have 6 vdevs
then it is possible to have 6 writes simultaneously occurring.
With raidz/2 each write must occur one at a time because the write
will occur on all disks in the raidz/2 vdev which has the consequence
of limiting the max iops to the speed of the slowest disk.
Neither exchange or SQL Server tend to be throughput bound, but
both require very high IOP/s rates. I don't think you'll see enough
traffic accross the controller for it to be the
bottleneck!<br>
Correct, there is not much traffic, but we have decreased iops for a
particular lun affected by activity on another volume (backup going,
defragmentation, on-line maintenance or something else, I'm not a
specialist here).
That would be true of any shared storage solution.
Just because it's ZFS doesn't mean you can now throw all workloads
together. You still need separate zpools for different workloads. If a
couple of workloads are slow sequentials, share a raidz/2 zpool of
SATA disks. If you have a high transactional system, give it a
dedicated zpool of raid10 and so on.
Even with a dedicated SAN and 15K/rpm drives, MS generally recommend
Raid-10 configurations for exchange. Raid-5/6, or RaidZ1/2 usually
doesn't give the IOP/s rates you need - although many people do
anyway.<br>
That's why I'd like to use SSD - to improve iops to a desirable level.
SSD helps, but doesn't eliminate the short comings of raidz/2 pool
which is, iops of single drive.
The SSD would have the same effect as if you had a mirror of 2 SATA
drives, iops wise, so if you really want to go this route why not save
a lot and just buy 2 big SATA drives plus an SSD for your exchange dbs?
How many users are there on your exchange server? I have a suspicion
that even if you move to zfs mirroring, you still might not get
enough
performance for exchange with that number of drives -
especially if you're putting other load on the system.<br>
Yes, there is "other load" and this is a problem as well (archiving
in our case).
That's why they want to separate exchange dbs from each other.
That means separate zpools.
Raidz/2 for archive dbs is fine, but a zpool of mirrors is really
needed for primary exchange dbs.
-Ross
_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss
