I'm basically doing the same thing you want to, just not the amount of data you are talking about. I have my 2008R2 VM connected to it's file share data via Iscsi initiator inside the VM. That way the volumes can be directly snapped on my EQ boxes without any concern for the VM itself. I've never had to use it though with shadow copies turned on. My PS5000 and 6000 are both configured for Raid 50. With the VMWare HA in place I don't really see the need for clustered storage, but your requirements may be different. I'm not doing MPIO on the file server, don't need to for the few hundred meg of data on it. But I am doing MPIO for my VM SQL Boxes. And soon Exchange. VMWare has a excellent white paper on best practices for Exchange 2010 running in a VM using Iscsi and MPIO.
-----Original Message----- From: Ian Roche [mailto:[email protected]] Sent: Thursday, October 28, 2010 10:02 AM To: NT System Admin Issues Subject: Virtualization of file server - storage recommendations. Hi Guys, I am about to undertake a new project which as the subject line states its turning our current file server into a vm. Its currently a clustered windows system (2 x nx1950's) connected via SAS to an md3000 + md1000. The quantity of data living on the storage is in or around 6TB (allot of this would be static) and its split over 2TB drives.The. We have a relatively new vmware environment consisting of 4 hosts (Dell R900's and M710 blades running ESXi4) they are connected to ISCSI Equallogic PS6000 units which are all in a RAID50 config. That's the hardware background, my query is around the virtualization of the storage for the virtual file server. I have read a number of article but would be interested if anyone has "real world" experience of what the feel works or definitely doesn't work. My plan is to have a small front end vm running win srv 2008 r2 ( just a single vm that will be protected by vmware HA) sitting in front of the storage . My question is how do you guys think its best presented I'm leaning towards standard vmfs datastores on dedicated 2tb luns purely being accessed by this new vm or using the equallogic mpio and the windows iscsi initiator to present the storage this way (which we have done on a couple of sql boxes already). From what I have read so far I don't see any particular reason to go down the RDM route other than some people saying if you have large drives the rdm path can be a good idea. I don't see a huge performance difference between vmfs and rdm in any of the documentation I have read to date. I would appreciate any comments or suggestions on this one its at an early stage but I just want to get the map down for the migration. Whatever route I choose the swap over will be done inline, I will migrate the data to the new vm (using robocopy) kill access to the physical box do a final diff and name the virtual system to the correct name (the fqdn is very important in my company it referenced in allot of build scripts as we are a software house). Cheers Ian ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ <http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/> ~ --- To manage subscriptions click here: http://lyris.sunbelt-software.com/read/my_forums/ or send an email to [email protected] with the body: unsubscribe ntsysadmin ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ <http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/> ~ --- To manage subscriptions click here: http://lyris.sunbelt-software.com/read/my_forums/ or send an email to [email protected] with the body: unsubscribe ntsysadmin
