From: | David Rees <drees76(at)gmail(dot)com> |
---|---|
To: | George Sexton <georges(at)mhsoftware(dot)com> |
Cc: | Greg Smith <greg(at)2ndquadrant(dot)com>, pgsql-performance(at)postgresql(dot)org |
Subject: | Re: 3ware trivia overload |
Date: | 2010-09-20 22:09:41 |
Message-ID: | AANLkTinX5Aeu_2O+PG2j8wrGecSf1rJV0HXvrxR+_5f7@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-performance |
On Mon, Sep 20, 2010 at 2:54 PM, George Sexton <georges(at)mhsoftware(dot)com> wrote:
> I'll throw in my 2 cents worth:
>
> 1) Performance using RAID 1 for reads sucks. You would expect throughput to
> double in this configuration, but it doesn't. That said, performance for
> RAID 1 is not noticeably worse than Linux MD. My testing showed the 3Ware
> controller to be about 20% faster than Linux MD for RAID 1.
No performance improvement is expected for streaming reads in any
non-striped RAID1. Random reads should nearly double in throughput.
You can use Linux's software RAID10 module to improve streaming reads
of a "mirror" using the "far" layout which in effect stripes the data
across 2 disks at the expense of some hit in streaming write
performance. Testing is required to determine if this tradeoff works
for your workload or not.
-Dave
From | Date | Subject | |
---|---|---|---|
Next Message | Kevin Grittner | 2010-09-20 22:28:53 | Re: Auto ANALYZE criteria |
Previous Message | George Sexton | 2010-09-20 21:54:32 | Re: 3ware trivia overload |