Storage dilemma - EMC -> HP
Hi all,
We currently run an EMC CX3-20 with the following spec:
120 x 300GB 15,000RPM FC disks
broken out to:
5 disks = vault pack
30 disks = RAID50
remainder = RAID10
Since the CX3-20 is limited to 120 disks (and we need more space for the RAID50 load), we need to either upgrade the CX to a CX3-40 (or CX4-240), or move the load (VMware Lab Manager) on RAID50 to alternative storage. For internal reasons we've decided to go with HP, but have a dilemma. The choice is:
HP P2000fc G3 with 36 300GB 15,000RPM SAS drives (2 x 2GB cache)
or
HP EVA4400 with 36 300GB 15,000RPM FC drives (2 x 2GB cache)
So, what we need to know is will the P2000 perform the same as the EVA4400 (IOPS/throughput), and how will that compare to the existing environment. From the specs, it looks like performance should be fine and the EVA only provides more management (and increased scalability), but the only two areas we are concerned about are:
1) Performance
2) Reliability
Since both the P2000 and EVA have redundant controllers, multipath I/O, etc, I don't think reliability will be an issue?
Any and all comments welcome, but we'd prefer technical explanations rather than marketing or gut feelings!
Thanks in advance!
S.
We currently run an EMC CX3-20 with the following spec:
120 x 300GB 15,000RPM FC disks
broken out to:
5 disks = vault pack
30 disks = RAID50
remainder = RAID10
Since the CX3-20 is limited to 120 disks (and we need more space for the RAID50 load), we need to either upgrade the CX to a CX3-40 (or CX4-240), or move the load (VMware Lab Manager) on RAID50 to alternative storage. For internal reasons we've decided to go with HP, but have a dilemma. The choice is:
HP P2000fc G3 with 36 300GB 15,000RPM SAS drives (2 x 2GB cache)
or
HP EVA4400 with 36 300GB 15,000RPM FC drives (2 x 2GB cache)
So, what we need to know is will the P2000 perform the same as the EVA4400 (IOPS/throughput), and how will that compare to the existing environment. From the specs, it looks like performance should be fine and the EVA only provides more management (and increased scalability), but the only two areas we are concerned about are:
1) Performance
2) Reliability
Since both the P2000 and EVA have redundant controllers, multipath I/O, etc, I don't think reliability will be an issue?
Any and all comments welcome, but we'd prefer technical explanations rather than marketing or gut feelings!
Thanks in advance!
S.
ASKER
Thanks simonlimon.
I really wanted a real-world experience view, if possible. Looking at the HP stats, they (naturally) don't provide figures for exactly the same tests on the two different systems. People generally quote IOPS based on disk performance, and both SAS and FC drives of the same rotational speed should perform similarly.
I concede that cache effectiveness and algorithms will make a difference (in some cases), but the bus and hardware design should have minimal effect since disk I/O is so much slower than system I/O.
Any comments appreciated!
Thanks!
S.
I really wanted a real-world experience view, if possible. Looking at the HP stats, they (naturally) don't provide figures for exactly the same tests on the two different systems. People generally quote IOPS based on disk performance, and both SAS and FC drives of the same rotational speed should perform similarly.
I concede that cache effectiveness and algorithms will make a difference (in some cases), but the bus and hardware design should have minimal effect since disk I/O is so much slower than system I/O.
Any comments appreciated!
Thanks!
S.
i am not sure that the EVA with 36 disks will give you the capacity you want!#
What kind of useable capacity do you want?
What kind of useable capacity do you want?
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Thanks all for the responses. We went with the MSA for now and will see how it goes. If the performance suffers we will adjust our loads..
I am a little disappointed with the 16 disk limitation on the vdisks though (for RAID10). That physically limits the performance to 16 disks for read (assuming it can read from raid mirrors simultaneously) and only 8 disks for write. Due to this we're not hopeful that the system will perform to the same level as before...
Thanks!
S.
I am a little disappointed with the 16 disk limitation on the vdisks though (for RAID10). That physically limits the performance to 16 disks for read (assuming it can read from raid mirrors simultaneously) and only 8 disks for write. Due to this we're not hopeful that the system will perform to the same level as before...
Thanks!
S.
But I believe EVA will provide better performance as it uses a different bus than the P2000.