When we look at SBS 2003 and the growth in product and features over its lifespan we have an idea of how the single box will perform.
SBS 2008 and Exchange 2007 we encountered an exponential growth in the need for disk I/O due to Exchange and RAM due to both Windows Server (Vista code) and Exchange.
Move to SBS 2010 and Exchange 2010 and if we were keeping an eye on the various product groups and their direction for the product we would have seen _before_ SBS 2011 STD ever RTMd that Exchange 2010 was designed to run on one SATA hard disk with everything in RAM. We would have then planned our deployments, both physical and then virtual as that became much more common, around the server products built-in.
The key to any single host design or cluster design is in what will be running on top of them. Obviously, but maybe not?
Here, our experience comes into play if we have been taking the SBS product and tearing it apart for the last ten years and three major product iterations. The inner-workings of SBS, Exchange, Active Directory, Group Policy, SharePoint, and so many other server feature sets were there for us to explore. Not only that, Microsoft gave us a really solid template to carry forward into our now stacked solution sets.
- VM0: AD, DNS, DHCP
- VM1: Exchange or LoB
- VM2: SQL or LoB
- VM3: RDS
Provisioning the above has not changed in a sense. We need to augment our host configuration for the extra 15GB of OS space per VM perhaps. But, for the most part our physical hardware will be similar in nature to what we would deploy for previous versions of SBS Standard _given the products running in the suite_.
The key in all of this is knowing how the various server products will behave given certain workloads.
SQL has an I/O tester. Exchange has a load tester called Jetstress. Those two utilities can help us understand what our small, medium, and large clients can expect for a given server topology. They can also help us to deliver a solution tailored to their specific needs.
Having a lab is key to getting to know the products and how to put them together.
Testing _every_ solution that goes out the door before actually setting up the client’s own solution set is also critical.
Knowledge is key to our value to our clients. Lose that and we’ve pretty much lost the game.
It takes lots of time. It can take a lot of money. But, in the end training and grinding away at configurations in a lab is key to our client’s success and to ours as well.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book
Chef de partie in the SMBKitchen
Find out more at
Windows Live Writer