On my Sun StorageTek 6140 array, I need to create two vdisks, one for a single NTFS file server volume and another for a single VMFS volume. The file server is run-of-the-mill, and the VMWare volume will be shared between four ESX 4.0 blades. I think I'm going to use RAID 6, which excludes me from using one of the default storage profiles (copied below), and I need some help figuring out how best to configure the profile for each purpose.
Is anyone aware of a good reference that can tell me, for instance, what segment sizes are good for what applications, file systems, etc.? I think I've figured out that 64KB is good for the NTFS vdisk and volume, but I really don't know what would be best for VMFS. Any recommendations?
Also, I do not have an available VMware initiator host type, per se. I should probably just use an ordinary Linux initiator, right?
Thanks!
This Question has been solved and asker verified All Experts Exchange premium technology solutions are available to subscription members.
Experts Exchange has been collecting answers to technology questions since 1996…3 million and counting! If you have a question, chances are we already have your answer.
If you can't find the exact answer you're looking for, ask our exclusive community of 50,000 experts. You’ll get a personalized answer from a trusted professional.
Thousands of free tech tips, tricks, how-to’s and tutorials are available in our peer reviewed articles section. See for yourself how smart our experts are, no login required.
Save solutions to your questions, answers you’ve discovered through searching plus helpful articles in your personal knowledgebase for easy future access.
Access the answers to your technology questions today.
30-day free trial. Register in 60 seconds.
Members of the expert community talk about why the experience at Experts Exchange is different than what you will find anywhere else.
Try it out and discover for yourself.
30-day free trial. Register in 60 seconds.
Join the community of experts here and help other tech pros by answering question in your area of expertise. You can earn FREE access to all Experts Exchange's premium features and resources.
Unfortunately, that's not an easy question to answer as some of those parameters depend on the array - but as rules of thumb:
- The more spindles (drives), the better!
- For a *random workload (Exchange, SQL, VMware)*: A 15K SCSI, SAS or FC drive can produce about 200 IOPS (I/Os per second) in a peaking to a little over twice that before performance falls off a cliff. Multiply the number of drives in the RAID group (including parity drive) by the IOPS and you have the performance of the RAID group.
- For a *random workload (Exchange, SQL, VMware)*: A 10K SCSI, SAS or FC drive can produce about 140 IOPS (I/Os per second) in a peaking to a little over twice that before performance falls off a cliff. Multiply the number of drives in the RAID group (including parity drive) by the IOPS and you have the performance of the RAID group.
- For a *random workload (Exchange, SQL, VMware)*: A SATA drive can produce about 60 - 80 IOPS (I/Os per second) in a peaking to a little over twice that before performance falls off a cliff. Multiply the number of drives in the RAID group (including parity drive) by the IOPS and you have the performance of the RAID group. The thing to note here is that SATA is *not* a suitable home for random workloads.
Parity RAID write penalties can bite you in the bum. For every random host write, a RAID 5 set must perform 4 disk operations: read data, read parity, calculate new parity, write new parity, write new data, so if you have a 2000 IOPS host workload, what you've actually got going on at the disk is (assume 50% read, 50% write) 1000 read + (1000*4) write = 5000 IOPS. If you calculated the number of disks on the host workload, you'd have 10 disks (2000/200). What you actually need is 25 disks (5000/200). If you only had 10 disks, they would be running at 500 IOPS per disk (well, they wouldn't - they'd be flat out) and your host would be hurting big. Conversely, a RAID 1/0 set only has two writes per host write (as the data must be mirrored), so the numbers of disks you need becomes 1000 R + (1000 * 2) W = 3000 IOPS. 3000 IOPS only needs 15 drives for the same host performance. So, contrary to what most people think, RAID 1/0 can actually save you money! The thing to be aware of is that, as drives become larger - and EMC has just announced 600GB FC drives, you can fulfill your space requirements with fewer physical disks, but you then won't have enough physical disks to fulfill the performance requirements. You need to do your homework diligently, or you'll end up with poorer performance in your SAN than if you'd left everything alone.
A useful tool to assess performance is VMware's Capacity Planner assessment, or you can use Windows perfmon or native -NIX tools.
>What additional performance analytics are used to determine optimal segment sizes?
Typically the file system helps determine that - but it also depends on the array and how it handles cache and RAID write optimisations. Segment size is more important for sequential operations where you cxan get some very helpful performance gains through aligning the size of the host write operations with the underlying stripe size. SAN performance tuning can be a pretty complex subject. :-)
Generally speaking, though, would you agree with this statement:
"Middle-of-the-r
In my case, the choices are 16K, 32K, 64K, 128K, 256K, and 512K. My best guess is that for a RAID 6, NTFS file server volume comprised of 8 x 300GB, 15K FC disks, 128K segments are the ticket. That same number/type of disks comprising a RAID 5 VMFS volume, however, would be better served almost universally by the largest available segment size, i.e. 512K.
Do the above statements and numbers sound like a reasonable basis from which to begin?
Business Accounts
Answer for Membership
by: meyersdPosted on 2009-09-03 at 22:36:40ID: 25257030
Use High_Performance_Computing