Storage carving recommendations?

We are purchasing our first SAN, a Dell MD3220 with 24x 146GB 15K 2.5" SAS disks.  Because I have little experience with SANs, I'm seeking input on the recommended configuration for these disks.

We have 200 desktops, all are virtualized and rely on Citrix PVS for reads.  Writes are redirected locally (and lost on reboot), except for profile changes, which are streamed to/from our file server (2008R2 Server Core) as needed.  Citrix recommends RAID 5/6 for PVS, but it seems like a better idea to use RAID 10 for my file server.

We also have a virtualized Exchange 2010Sp1 setup for 120 users, but only concurrent usage of ~ 75.  Our AD VM also holds data for about 120 users, as well as providing DNS/DHCP/NPS for all 200 users.  It seems like RAID 10 is the best idea here, as well.  

We also run about 10 other VMs, including 2 XenApp VMs (1 for hosting, 1 for streaming/management), vCenter (we are 100% virtualized on servers and desktops) for 3 hosts, and small application servers.

We do use PerfectDisk to defrag all our VMs.

It is my understanding that disk groups are assigned to controllers, and in this regard, it seems like my best option to get maximum performance is to create a RAID 10 group for Exchange, AD,  file server and XA published, as well as a RAID 5/6 (we have a cold spare) for Citrix PVS, XA streamed, and the rest of my small application servers.  Each group would be assigned to its' own controller.

Thoughts/suggestions/comments?  One idea I had was to disable the read cache on the MD3220, dedicate all cache to writes, and increase RAM in each my VMs to utilize Windows' caching (we are around 25% RAM utilization on all 3 hosts, <8% CPU).  

We will not be using SAN snapshots or virtual disk copies, but may purchase the performance software if budget allows.
Who is Participating?

[Product update] Infrastructure Analysis Tool is now available with Business Accounts.Learn More

I wear a lot of hats...

"The solutions and answers provided on Experts Exchange have been extremely helpful to me over the last few years. I wear a lot of hats - Developer, Database Administrator, Help Desk, etc., so I know a lot of things but not a lot about one thing. Experts Exchange gives me answers from people who do know a lot about one thing, in a easy to use platform." -Todd S.

Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
Personally, I would create one large RAID 10 Array for all your disks.

and then carve out LUNs, of between 500 - 800GB, to present to the VMware Host ESX Servers. If those LUNs are large enough to support your Virtual Disks.

Experts Exchange Solution brought to you by

Your issues matter to us.

Facing a tech roadblock? Get the help and guidance you need from experienced professionals who care. Ask your question anytime, anywhere, with no hassle.

Start your 7-day free trial
sbumpasAuthor Commented:
I considered that, but I have 2 questions:

Are LUNs assigned to controllers?

Is it true that storage slows down significantly as you surpass the 50% capacity threshold?
whomever gave you that arbitrary 50 percent rule is wrong.  storage slows down as activity increases and buffer / write cache becomes less effective.

the  best way to configure storage is to minimize unnecessary I/O and  take advantaage of performance characteristics of different RAID levels.  mix and match.  use RAID1 for the most random IO.  scratch table space and log files. Raid6 (RAID5 is unacceptable IMO due to risk of data loss WHEN, NOT IF you lose a drive.  one bad block will tyically result in MBs of lost data, more if it is in a database file)

if you want screaming performance, buy 2 small SSDs &  mirror them .  you can get 50000  random I/Os per second vs a few hundred with all your remaining storage put together.  slice and dice the heck out of it and make that the database indexing workhorse.  remember vmware is going to be writing 512KB or more if you so much as touch a file if data is on R5 OR R6.  RAID1 doesn't have that overhead. that makes RAID1 even more advantageous.  R10 also guarantees unnecessary I/O but not as bad as R5/6.

with that many disks you may have bus saturation unless your exander is dual ported and you use SAS instead of SATA
Determine the Perfect Price for Your IT Services

Do you wonder if your IT business is truly profitable or if you should raise your prices? Learn how to calculate your overhead burden with our free interactive tool and use it to determine the right price for your IT services. Download your free eBook now!

sbumpasAuthor Commented:
Unfortunately, 2 small SSDs from Dell cost much more than is reasonable.
Are AD and roaming profiles considered random IO?  I know Exchange certainly is.
Dell doesn't make SSDs, so you're paying a huge premium to begin with.  Just buy two fewer HDDs and a pair of OCZs or something similar from an online retailer who can bundle a low-end RAID1 controller that works with it and VMWARE. Reserve the SSD for SQL Index files and scratch table space. You will be limited on throughput, but not on I/Os per second.  The roaming profiles are random, but Exchange is more important.
sbumpasAuthor Commented:
Doe anyone know where to find a storage calculator (performance based) for file servers/roaming profiles, similar to the Exchange sizing calculator?
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
Probably better asked as a new Question, as that question is hidden in this Question, it will not get the exposure you require.
sbumpasAuthor Commented:
Thanks for everyone's input!
It's more than this solution.Get answers and train to solve all your tech problems - anytime, anywhere.Try it for free Edge Out The Competitionfor your dream job with proven skills and certifications.Get started today Stand Outas the employee with proven skills.Start learning today for free Move Your Career Forwardwith certification training in the latest technologies.Start your trial today

From novice to tech pro — start learning today.