Exchange 2010 DAG on local disk or iscsi.
Can anyone please give me your thoughts on the following design questions and recommendations. We have NetApp with NFS so we have limited choices on how we can implement this. Option 1. VMware-ESX Host 1 VM-Exchange1 Storage-NFS 2 Active DAGs and 2 passive DAGs all on local disk of the VM. VMware-ESX Host 2 VM-Exchange2 Storage-NFS 2 Active DAGs and 2 passive DAGs all on local disk of the VM. Setup replication service to replicate Active to Passive DAGs on each Exchange server. Disable HA and DRS on vCenter. Option 2. VMware-ESX Host 1 VM-Exchange1 Storage-NFS 2 Active DAGs using ISCSI software initiator. VMware-ESX Host 2 VM-Exchange2 Storage-NFS 2 Active DAGs using ISCSI software initiator. Storage will be centralized with Microsoft Clustering enabled, both Exchange servers will have Active/Active roles but are capable of hosting all 4 DAG’s if needed. Disable HA and DRS on vCenter.Jeff D
January 14th, 2011 11:50pm

First, I think you mean 4 database copies per server; Not 4 DAGs per server. Anyway, it sounds like your question really is: Should I present iSCSI to the Guest OS via VMware or directly via initiator? Right? This is a good question to ask. I suppose I’d consider the overall environment first. Is the same SAN presenting storage the storage used by VMware for the guests themselves? If so, it seems like it’d be easier to have it all going to the same place for a more consistent failure, should one occur. However performance might be better if you send the storage to the guest directly. Also, this allows Exchange admins to troubleshoot storage without involving the VMware folks. All things being equal, I’d probably present the storage to the guests directly via their own initiator. Finally, since we’re talking about DAGs on VM’s it’s prudent to point out virtual machine clustering technology and DAGs are not supported. You didn’t say you were doing this, but I wanted to mention it: >>Microsoft doesn't support combining Exchange high availability solutions (database availability groups (DAGs)) with hypervisor-based clustering, high availability, or migration solutions that will move or automatically failover mailbox servers that are members of a DAG between clustered root servers. DAGs are supported in hardware virtualization environments provided that the virtualization environment doesn't employ clustered root servers, or the clustered root servers have been configured to never failover or automatically move mailbox servers that are members of a DAG to another root server. << -src: http://technet.microsoft.com/en-us/library/aa996719.aspx --------------------------------------------------------------------------------
Free Windows Admin Tool Kit Click here and download it now
January 15th, 2011 7:26pm

I am in the same boat. VMware 4.1 with NetApp pure NFS. Currently we run Exchange 2003 with the C and D drives on VMDK (vmware) and because we use NetApp Snap Manager for Exchange we have iSCSI (Same SAN) LUN's for the log and db drives (L and M in our case). We are moving to Exchange 2010 and we are ditching the SnapManger for Exchange and using DPM 2010. I am going to run all of now on VMDK/NFS and drop iSCSI. This is what we did with SQL in VMware/NFS and it works great over 10gb/Nexus.Thanks, -Lindy
January 17th, 2011 7:49am

Thank you for the clarification, To elaborate some more, what is your opinion if we took out iSCSI completely out of the picture and just stored the databases on local disk which would be the VMDK files. VMDK files are NFS to NetApp. We would then have Active databases in 1st datacenter and Passive databases in 2nd datacenter for DR/Failover. I have not tested but I believe the passive databases are replicated up to the minute? When you mention clustering is not supported, we wouldn't be using the traditional MS clustering but doesn't Active/Passive DAG's in different Datacenters setup a form of clustering within Exchange? CWLindy, I believe you are doing what we are planning to do and present the MS Exchange databases on local disk such as: C:\OS.vmdk D:\DB.vmdk E: \Llogs.vmdk ETC Jeff D
Free Windows Admin Tool Kit Click here and download it now
January 18th, 2011 6:45pm

A few things to note. DAG are not supported if you use VMware DRS (vmotion) so you have to PIN your Exchange Server VM's to a host and not let it move on its own. I have read many people say it still works. Also I have read that Exchange Databases on NFS, even via a virtual disk (VMDK) is not supported. However VMware ( or NetApp cant remember) posted some article that showed Exchange 2010 performance with FC, iSCSI and NFS and it was very good. All that said I will probably pin my server but still use NFS. VMware + NFS on 10gig is a very powerful setup that is very simple to administer.Thanks, -Lindy
January 18th, 2011 7:18pm

Hi Deploying DB's on NFS is not supported but I don't think there are anything about to deploy it using VMDK files, that should be OK if they are using fixed size and not dynamic expandable The storage used by the Exchange guest machine for storage of Exchange data (for example, mailbox databases or Hub transport queues) can be virtual storage of a fixed size (for example, fixed virtual hard disks (VHDs) in a Hyper-V environment), SCSI pass-through storage, or Internet SCSI (iSCSI) storage. Pass-through storage is storage that's configured at the host level and dedicated to one guest machine. All storage used by an Exchange guest machine for storage of Exchange data must be block-level storage because Exchange 2010 doesn't support the use of network attached storage (NAS) volumes. Also, NAS storage that's presented to the guest as block-level storage via the hypervisor isn't supported. The following virtual disk requirements apply for volumes used to store Exchange data: Virtual disks that dynamically expand aren't supported by Exchange. Virtual disks that use differencing or delta mechanisms (such as Hyper-V's differencing VHDs or snapshots) aren't supported. Note: In a Hyper-V environment, each fixed VHD must be less than 2,040 GB. For supported third-party hypervisors, check with the manufacturer to see if any disk size limitations exist. http://technet.microsoft.com/en-us/library/aa996719.aspx Jonas Andersson MCTS: Microsoft Exchange Server 2007/2010 | MCITP: EMA 2007/2010 | Blog: http://www.testlabs.se/blog | Follow me on twitter: jonand82
Free Windows Admin Tool Kit Click here and download it now
January 21st, 2011 2:04am

It is right here..... http://technet.microsoft.com/en-us/library/aa996719.aspx "All storage used by an Exchange guest machine for storage of Exchange data must be block-level storage because Exchange 2010 doesn't support the use of network attached storage (NAS) volumes. Also, NAS storage that's presented to the guest as block-level storage via the hypervisor isn't supported." NFS/CIFS is file based storage. Fiber Channel and iSCSI are block based storage as in LUN. LUN's are just fixed containers or block based storage. In VMware it is super easy to connect to a SAN volume with NFS shares, it takes seconds to setup on the SAN and in VMware. That volume is file based. The VMware guest OS, say Windows 2008 R2, is siting on that NFS share, file based storage. VMware presents that storage as DASD to the the guest OS, and block based DASD at that, but its still File based storage underneath the virtual layer. No matter I am still going to use it. We have had Exchange 2003 with 1800 mailboxes running on VMware/NFS for over two years. I storage VMotioned it from a FC connected EMC SAN volume to a NFS share off of a NetApp SAN, live during the day. Try that with Hyper V.Thanks, -Lindy
January 21st, 2011 6:17am

I am running Exchange 2007 with my Mailbox server and Hub Transport/Client Access server running under ESXi 4.1 using NFS storage off a Netapp. I present to the Mailbox server two luns, one for the Exchange database and one for logs. I am migrating to Exchange 2010 and would rather present to the mailbox server two NFS shares. It appears that a few people might be running Exchange 2010 like this even though it is not a supported configuration. How is it running? Any gotchas? I have a small environment with 600 mailboxes
Free Windows Admin Tool Kit Click here and download it now
April 23rd, 2011 9:00pm

@HendersonD I dont know of anyone that uses Exchange 2010 on NFS in the way you are suggesting. (Windows server physical or virtual mounting NFS shares off a san to use for Exchange). With VMware you mount NFS shares from the VMware host or vCenter. The Host sees the disk space via the NFS share. Guest VM's that use that NFS disk space, just think they have direct attached storage, or DASD. You can do things such as RDM's or use iSCSI from a VM guest to jump out of the Hyper Visior and mount iSCSI LUN's to the VM Gues running Exchange. Thanks, -Lindy
April 23rd, 2011 11:25pm

Lindy, What is your Exchange 2010 setup? I have a SQL 2008 R2 server that runs under VMWare with the storage being NFS on a Netapp filer. This VM just has three disks: - C drive which is on the same NFS share as all the rest of my VMs - E drive, for this I carved out another NFS share on my filer and then just added another disk to the VM choosing this NFS share for storage. This is used for my SQL databases - F drive, this is done in a similar fashion as the E drive but it is used for SQL log files I am not mounting shares directly off of my SAN to Exchange, these are VMDKs stored on NFS shares. My thought was to setup my Exchange 2010 server in a similar fashion, your thoughts? Dave
Free Windows Admin Tool Kit Click here and download it now
April 24th, 2011 12:13am

Dave, We are on the same page. NetApp Volume, NFS share that points to the volume, mounted to ESXi hosts, Exchange 2010 vmdk disk files created on that NFS volume. While it is not supported, it works just fine. Here is a good VMware doc. http://www.vmware.com/files/pdf/vsphere_perf_exchange-storage-protocols.pdfThanks, -Lindy
April 24th, 2011 1:27am

Lindy, Have there been any issues with your using VMDKs stored on NFS for your Exchange 2010 database and log files? I know I rarely call Microsoft for support and I have never called them about support on my Exchange 2007 environment. I only have about 600 mailboxes Are you also running DAGs with VMWare HA enabled? I know this is not a supported setup so I am thinking about not using DAGs so I can enable HA Dave
Free Windows Admin Tool Kit Click here and download it now
April 24th, 2011 6:57pm

Dave, I dont use a DAG. We only have about 1300 mailboxes max and of that about 200 are more than light users. The rest are remote users in retail based jobs. I use Snapmanager VI for up to 5 days of snapshots for a VM's on production volumes and then for specific VM Windows servers, Exchange being one of them I use DPM 2010, which is a VM as well. The DAG concept is really nice, especially compared to the cluster options of Exchange 2007 and below. If we had more users or email was a higher priority or we had multiple sites that housed Exchange servers then I would consider it. We ran Exchange 2003 on FC EMC with VMware and I did a storage VMotion over to the NetApp on NFS volumes, during the middle of the day when migrating to the NetApp 2 years ago:) For 2003 or 2010 I have not had to call Microsoft for anything Exchange related. My DRS is it to fully automated so my Exchange 2010 (no DAG) probably moves around all the time. This last week I was taking down each of the 8 server in the VMware cluster to bump the RAM from 72gig - 96gig, one server at a time so Exchange has probably moved many times in the last week. Lastly I Dedupe the volumes that the Exchange DB vmdk files are on. With Exchange 2010 you lose SIS (single instace storage) because of the DAG stuff, so Dedupe works better now. With 2003 I did not get much out of dedupe on the DB volume because SIS was doing that for me.Thanks, -Lindy
April 25th, 2011 3:51am

Lindy, A quick follow up. Over the past week I stood up three virtual servers under Vsphere all running Server 2008 R2. These three will become my mailbox, hub transport, and client access servers for Exchange 2010. The only thing I have not done yet is provide storage on the mailbox server to store the database and log files. I can either add two iSCSI luns or two more disks based on NFS for this purpose. I have a consultant coming on campus this Monday to do the install/config of Exchange 2010 so I need to make a decision in the next few days. I am leaning towards just adding two disks to the VM based on two NFS shares off my netapp. As I mentioned before we are moving away from Snapmanager for Exchange and will be using Commvault to backup Exchange which is about 600 mailboxes and a 150GB datastore. Two quick questions: 1. How long have you been running Exchange 2010 using NFS storage for the database and log files? 2. Everything been running fine? The dedup piece is interesting. Right now I just dedup the volume I using to store my VMs and my CIFS shares. Once I switch to Exchange 2010 I may try it on the volumes holding the database and log files. Dave
Free Windows Admin Tool Kit Click here and download it now
May 10th, 2011 10:25pm

"Once I switch to Exchange 2010 I may try it on the volumes holding the database" You'll be pleasantly surprised. J
May 10th, 2011 11:52pm

This topic is archived. No further replies will be accepted.

Other recent topics Other recent topics