Thread: local-storage versus SAN sequential read performance comparison
This is outside of PG performance proper, but while testing pg_dump and pg_restore performance on local-storage versus SAN storage I noticed a big performance drops on in the SAN configuration. I'm told that both local storage and SAN storage are 15k drives with local-storage running dual RAID1 configuration while HP StorageWorks SB40c SAN is 10x15k RAID1+0. Running a hdparm -tT tests with different read-ahead, I see the following differences on /dev/sda (local-storage) and /dev/sdc (SAN storage). I'm shocked at the drop in buffered disk read performance 150MB/sec versus 80MB/sec and surprised at the SAN variability at 1MB/sec versus 10MB/sec, local-storage and SAN storage respectively.
For those who, unlike me, have experience looking at SAN storage performance, is the drop in buffered disk reads and large variability the expected cost of centralized remote storage in SANs with fiber-channel communication, SAN fail-over, etc.
If you have any ideas or insights and/or if you know of a better suited forum for this question I'd sure appreciate the feedback.
Cheers,
Jan
Attachment
How fast is the link to your SAN? If it's Gigabit, then 80MB/s would be pretty reasonable. Also SANs are normally known for very good random access performance and not necessarily for fast sequential performance. On Mon, May 21, 2012 at 8:17 AM, Jan Nielsen <jan.sture.nielsen@gmail.com> wrote: > This is outside of PG performance proper, but while testing pg_dump and > pg_restore performance on local-storage versus SAN storage I noticed a big > performance drops on in the SAN configuration. I'm told that both local > storage and SAN storage are 15k drives with local-storage running dual RAID1 > configuration while HP StorageWorks SB40c SAN is 10x15k RAID1+0. Running a > hdparm -tT tests with different read-ahead, I see the following differences > on /dev/sda (local-storage) and /dev/sdc (SAN storage). I'm shocked at the > drop in buffered disk read performance 150MB/sec versus 80MB/sec and > surprised at the SAN variability at 1MB/sec versus 10MB/sec, local-storage > and SAN storage respectively. > > For those who, unlike me, have experience looking at SAN storage > performance, is the drop in buffered disk reads and large variability the > expected cost of centralized remote storage in SANs with fiber-channel > communication, SAN fail-over, etc. > > If you have any ideas or insights and/or if you know of a better suited > forum for this question I'd sure appreciate the feedback. > > > Cheers, > > Jan > > > -- > Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org) > To make changes to your subscription: > http://www.postgresql.org/mailpref/pgsql-performance > -- To understand recursion, one must first understand recursion.
On 05/21/2012 10:41 AM, Scott Marlowe wrote: > How fast is the link to your SAN? If it's Gigabit, then 80MB/s would > be pretty reasonable. Wow, I didn't even think of that. I was looking up specs on the SAN. :) Those stats are pretty horrible for sequential reads. The SAN should improve random reads, not mercilessly ruin sequential. That does sound like the problem, though. -- Shaun Thomas OptionsHouse | 141 W. Jackson Blvd. | Suite 500 | Chicago IL, 60604 312-444-8534 sthomas@optionshouse.com ______________________________________________ See http://www.peak6.com/email_disclaimer/ for terms and conditions related to this email