Thursday, 22 July 2010

Intel Modular Server – Promise VTrak e310sD Has Arrived

A client that is running a three node Hyper-V Server 2008 R2 cluster on an Intel Modular Server (IMS) needs to have _a lot_ more storage configured for their highly available VMs.

Their current SPF file server is dying a harsh death and causing all sorts of problems with folder redirection and share availability.

In our initial investigations into augmenting storage for servers and/or clusters running on the IMS we ended up focusing in on the Promise VTrak series of direct attached storage units as indicated in the above blog post.

Well, our first VTrak unit, an e310sD 12 bay with dual controllers,  has arrived here at the shop. We are waiting on the 12 300GB 15K.7 Seagate SAS drives and the two Adaptec external mSAS cables that should be showing up tomorrow or early next week.

Once we have everything together, we will look to getting our own IMS fired up ASAP as we will be testing a few configurations on our own systems before looking to deploy at the client’s site.

Client IMS Storage Configuration

The client IMS in question is configured similarly to ours in that we have all 14 drive bays in the IMS configured with 146GB 15K Seagate Savvio SAS drives.

  1. Storage Group 1: 2 physical disks – Low I/O Requirement
    1. H-V 2K8 R2 OS volumes for nodes
    2. Single volume for H-V configuration files (VM memory files etc).
  2. Storage Group 2: 4 physical disks – Medium I/O Requirement
    1. TS VM Remote Apps installation and other data volume
    2. SBS VM secondary storage
    3. Desktop OS VM volumes.
  3. Storage Group 3: 6 physical disks – High I/O Requirement
    1. SBS VM OS, Exchange, and other volumes
    2. SQL VM OS and database volume
    3. TS VM OS volume

The Promise VTrak will be used to add some high capacity medium I/O performance storage for the cluster to allow for the company’s redirected folders as well as folder shares to reside as an attached volume within the highly available SBS 2008 OS VM.

Hyper-V VM IDE Limitation

Keeping in mind that we only have 4 IDE connections to work with in a Hyper-V based VM, we usually only have 2 VHDs IDE connected at a time in a production scenario.

One IDE connection hosts the VHD OS volume, one hosts the VHD with data sets, and one IDE connection remains open.

If there is a need to recover the VM, we can attach a remote VHD to the open IDE connection and it will be seen by the backup software as a source to restore from.

This configuration leaves one IDE connection left over for the Hyper-V “optical” drive that is always connected.

VM Backup Destination VHD

Note that the remotely accessed VHD is actually connected via the Hyper-V SCSI bus which is hot swap capable in Hyper-V Server 2008 R2 thus allowing us to “rotate” the backup VHDs. Backups to those VHDs are encrypted using ShadowProtect.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*Our original iMac was stolen (previous blog post). We now have a new MacBook Pro courtesy of Vlad Mazek, owner of OWN.

Windows Live Writer

3 comments:

Anonymous said...

Hi, I have a setup just similar with this one ... but I'm getting nowhere with accesing the LUNs on Vtrak(probably my fault, still whould be nice to understand where am I failing ...)
Normal full windows 2008 R2 installed servers in IMS can see and use the LUNs, but Hyper-V 2008 R2 servers can't ... neither single nor multi-server presented LUNs. AFAIK, all's well with fw's and drivers & co, the Vtrak controllers just never see the Hyper-V server's initiators showing up. Manually adding them got me nowhere.
If you get a working environement, I'd appreciate the details.
I'll be returnig here to check your progress, if you keep posting them.
Thanks in advance.
You can get me by email through ricardo dot santos at ra-csi dot com / same goes for IM/Live

Anonymous said...

Hi again,

Got the issue: cluster servers were not configured to access the SAS channel that connects to the E310sD.
Dumb and basic issue :-X
All's well and ok.

Thanks and good luck

Philip Elder Cluster MVP said...

A,

Thanks for that.

It has been a bit of a tough week with a migration and other client needs being ongoing so I have not been able to get to the VTrak yet.

Philip