Welcome to my first post in the forums! Rather than answering a question, however, I'm going to pose a question to the community to help answer a question for someone moving from Exchange '07 to Exchange '10.
The organization in question currently runs Exchange 2007 on a single server with all mailboxes stored on the SAN. They know that the design does not meet Microsoft official standards, but traffic patterns have meant that the scenario has worked very well for them.
With Exchange 2010 on the scene, they have a dilemma. Do they continue down the single-server path or do they implement a highly available architecture bases on Microsoft's services, such as DAGs and NLB? The organization wants a highly available system, but is wondering if VMware's own HA tools would provide enough of a benefit. In other words, do Microsoft's tools provide enough beyond ESX's tools to justify the additional resource needs - RAM, processor, etc., or are VMware's availability mechanisms good enough?
Further, if they want some semblence of scalability, would it make sense to start with a single mailbox server and two CAS/Hub servers?
The organization in question has about 1500 mailboxes on Exchange '07. About 250 of those are faculty and staff and the rest are students. Of those 250 faculty and staff, 70 or so are not on for good chunks of the day. Of the student mailboxes, most use OWA to check mail and are not in and out of mail all the time. As such, IO load does not really fit MS typical profiles.
Posts: 36
Joined: 29.Jun.2010
From: VA
Status: offline
Hi Scott, I don't have an answer to your question just yet, I'm researching the same question myself at the moment. I notice it's been over a month since you posted this, have you found any recent documentation or guides so far?
Not yet, and my trip to TechEd muddied the waters even more as some folks said that using RAID with DAGs ALMOST doesn't make sense since RAID (esp. RAID 10) is another copy anyway, so then what's the point of a DAG? I understand what they're saying, but need to kick around all of the options and then go forward. I'm not sure how uch I agree with the RAID/DAG argument.
I'll definitely report back once I have more information and pick a direction... probably next week.
i have plans to go from EX2007 to EX2010 very soon and i`m discussing with my boss about the VMware option to virtualize our Exchange Organization, just 400 users over here.
I hope Scott report some progress so we can get a better direction about this.
Posts: 36
Joined: 29.Jun.2010
From: VA
Status: offline
Interesting article, thanks for posting! I'm not sure how much I agree with going RAID-less either Scott, even if not technically necessary it would seem that you're putting yourself at unnecessary risk.
I'm planning and testing configurations for what will be a very large Exchange 2010 deployment. From what I've read a DAG will function just fine in an HA/DRS cluster, but I'm mostly concerned with supportability at the moment. My understanding is that MS won't officially support mailbox server VMs configured in a DAG if the host servers are configured in any kind of HA cluster. That is not a good thing in VMware terms since I'd rather not leave a couple ESX hosts out of the cluster just to support mailbox servers. They also say they don't support vMotion (or Live Migration to be fair).
I'm looking forward to anything Scott can find out as well.
Posts: 36
Joined: 29.Jun.2010
From: VA
Status: offline
As an update, MS seems to have changed thier tune a bit when it comes to the host servers being part of an HA/DRS cluster. Here's a blurb from technet regarding Exchange 2010 virtualization:
"
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."