Unsolved
This post is more than 5 years old
22 Posts
0
1128
Performance waring in Oracle when migrated to VMAX from DMX-3
We have an Oracle DB running on a Solaris server that we recently moved off a DMX-3 to a VMAX array and Oracle immediately began spitting out this warning.
"The throughput of the I/O subsystem was significantly lower than expected."
The only configuration difference here between DMX and VMAX is that we are using Virtual Pooling on VMAX. Does anyone know if there are problems with Oracle DBs and Virtual Pools? Thanks
KeystoneCruiser
22 Posts
0
August 24th, 2011 16:00
FAST is not enabled. The drives are 450GB FC (roughly 560 drives) in one VP pool (RAID5 3+1).
MorbidAngel1
62 Posts
0
August 24th, 2011 16:00
A lot of considerations required.What are the disk technology used in the Pool...FAST is enabled on the pool or not....if EFD,FC,SATA are used in the Virtual pool and FAST VP is enabled it should not be a problem.....Can you share the Virtual Pool Information???
MorbidAngel1
62 Posts
0
August 24th, 2011 16:00
Only FC Drives???NO EFD's??
Quincy561
1.3K Posts
0
August 24th, 2011 18:00
Same front end configuration? FAs? Same configuration of meta volumes if you were using them before?
dynamox
20.4K Posts
0
August 24th, 2011 18:00
that's a pretty nice, wide spread. Were devices configured as fully pre-allocated ? Do you have tools like SPA or Performance Manager to tell you what's going on in the array ?
KeystoneCruiser
22 Posts
0
August 24th, 2011 20:00
The devices were configured as fully allocated. We used PPME to move the data so the devices/blocks are the same. The FA configuration is also the same. The FAs are sharing with more servers but the iops/throughput is comparable to traffic on the DMX. SPA shows that the Port Group is only 15% utilized and the array appears to be in ideal shape. IOPS for the FAs are on average about 1,250 iops.
MDSchneider
25 Posts
0
August 26th, 2011 11:00
What's the response time for the storage group in SPA?
SKT2
1.3K Posts
0
August 26th, 2011 12:00
You have posted a very interesting message from the Oracle. Where do you see this message in Oracle?Do they have a performance issue in real? Ask your DBA to get more details using thier Average Waittime Report and then you can go from there
AJK3
108 Posts
0
August 29th, 2011 12:00
And also can check if there are any IO waits being seen. Is ASM used?
KeystoneCruiser
22 Posts
0
August 29th, 2011 17:00
The average read response time is 6ms and the average write time is 1 ms. The WP counts are minimal, the spike during DB backups but fall quickly to 0-100. The DBA is not reporting any DB latecy, just a concerning message as we are scheduled to move the producation DB's soon.
RobertDudley
448 Posts
0
August 30th, 2011 06:00
The problem may not be on the array. Do you have any internal disks that are used for TEMP or logging that may be a bottleneck?
5449Pr4QlS12567
46 Posts
0
August 30th, 2011 09:00
Also check for any new Deamon process you might have turned on. U might wanna check your File System usage.