The past couple of weeks have been a combination of flights, hotels, late-night meet-and-greets, early morning sessions, and overall mind exhaustion. It is conference time in the tech world, and while the sessions I've been speaking about and attending are diverse, there is an overall theme that seems to be running throughout: consolidation, saving money, and going green.
Optimizing solution environments is more critical now than ever. Challenging economic times have led to reduced budgets and the need to do more with less. And while some initiatives are being delayed or cut, it actually may make sense to move forward with others. I know spending money or making changes seems like a huge mistake, but there are times when it may benefit your organization to do so. Let's take Exchange 2007, for example.
[ For the latest virtualization news and views, check out David Marshall's Virtualization Report blog. ]
If like most, your Exchange infrastructure is distributed and you're on Exchange 2000 or 2003, or a mix, moving to the more scalable and feature-rich Exchange 2007 can actually save money. True, moving to Exchange Server 2007 requires x64 servers, but with the aid of consolidation methodologies and virtualization, you may be able to reduce your server count by up to two-thirds.
While virtualization is fast becoming common across the IT landscape, some are hesitant to deploy it in mission-critical environments such as e-mail. More and more organizations are moving past this risk aversion and have been demonstrating that it is indeed a viable option that can pay off handsomely. Getting rid of aging and expensive-to-maintain 32-bit servers can be a boon to the bottom line, and replacing them with far fewer x64 servers can further reduce expensive datacenter real estate, software licensing, power and cooling costs, connectivity costs, and management expense.
Virtualization's potential extreme consolidation
At the Exchange Connections event in Orlando two weeks ago, the solution vendor Unisys articulated its approach to Exchange -- Consolidated Exchange Solution (CES) -- which relies on a highly tested architecture for enterprise-class mission-critical Exchange. Using consolidation methodologies, including virtualization (both VMware's ESX and Microsoft's Hyper-V), Unisys has demonstrated impressive reductions in e-mail environments' footprint and cost while increasing service level and functionality. One of the customers discussed in the Unisys session reduced its number of e-mail servers from 60 to two, each with three hard partitions, while increasing the number of mailboxes by 33 percent.
There is quite a bit of back and forth on the savings one might achieve with consolidation and virtualization, but no doubt a reduction from 60 servers to two is a monumental accomplishment (granted they were monster servers, ES7000s with 24 dual cores and a boatload of memory comparable to six standard form factor, four-socket quad-cores).
Virtualization gotchas to note
There are some aspects to virtualization that you should be aware of before you proceed with your Exchange environment -- or any server environment. Make sure you decide what you will consolidate and what you will virtualize. There is a difference (although some don't realize that). One reason for the overwhelming server sprawl is due to deployment strategies in times past that encouraged a server in every location (leading to tremendous server underutilization). Now we see that server consolidation in fewer key sites provides for better server density, higher server utilization, fewer servers to manage, less attack surface, lower costs, and the most coveted of them all: green IT.
When considering virtualization, you need to think about the method you will use. Will you virtualize your physical environment into a virtualized environment, or will you use virtualization for a disaster recovery product? So you need to think physical to virtual, virtual to virtual, physical to virtual to virtual -- it can get a bit cumbersome. You need to decide the type of virtualization software you will use (VMware's ESX and Microsoft's Hyper-V being key players these days).
You also need to know what you can and cannot virtualize. Unified Messaging server for Exchange and the Office Communications Server are two examples of servers that are not supported by Microsoft for hardware virtualization. That isn't to say they won't work, but you won't have support if there is a problem.
Other Exchange roles like the Hub Transport and Client Access Server roles are perfect for virtualization. The Mailbox role may be a bit of a debatable issue in terms of performance losses for virtualizing. On one hand, you may lose some performance off the server by putting your Mailbox servers in as a virualized box with other systems, which may be counterproductive to your goal of a high-performance Exchange environment. However, you might consider this approach if you have an extremely large mailbox environment: There is a hard limit on connections for physical Mailbox Server instances.
While you may or may not decide to virtualize your production environment, you may consider using virtualization for disaster recovery. Granted, from a cost perspective you might use Cluster Continuous Replications (CCR) to create a failover solution on-site and then use Standby Continuous Replication (SCR) for the passive off-site disaster recovery. Or you might consider other, more costly software solutions like CA's XOSoft High Availability or Unisys' SafeGuard hardware replication appliance. What are the benefits to these solutions over out-of-the-box Cluster Continuous Replication (CCR), Single Copy Cluster (SCC) and Standby Continuous Replications (SCR)? Well, one thing is a rewind feature for more granular recovery from a failure. Of course, such capabilities come at a price.
So, I ask you, InfoWorld readers, what methods and solutions are you using to consolidate, save money, and go green? And what would you like to read more of on the Enterprise Windows front?