On May 7 I wrote about the curious lack of planning tools for Exchange 2013. One week later, Microsoft released Version 5.1 of the Exchange 2013 Server Role calculator with an accompanying explanation from Ross Smith IV posted on EHLO.
I’m still curious. First, why did Microsoft not release the calculator when they published the performance information for Exchange 2013 on May 6? Would it not have made more sense to release the two together? Perhaps it’s just that Microsoft wanted us all to read and educate ourselves from the performance information released a week earlier so that we would understand all of the complicated calculations that the macro-laden spreadsheet (E2013Calc5.1) spits out. The cynic in me says “no” and that the phased release was simply marketing doing its best to get maximum publicity for the release of tools that should really have appeared some time ago.
The second curious element is the version number. Where does 5.1 come from? After all, the current version of the Exchange 2010 mailbox server calculator is 19.9. All manner of intriguing possibilities come to mind to explain why the version number isn’t a sensible and logical 1.0 – or even 2.0. But version five? And a minor version number added for good measure? Madness.
But moving to a more positive perspective, it is great to see the Exchange 2013 calculator appear. It’s also good that Microsoft has not simply rested on their Exchange 2010 laurels and just updated the calculator to take account of changes in the areas of the Store and DAG such as the reduction in the number of databases that can be mounted on a mailbox server and different kinds of DAG layouts across multiple datacenters; instead, they have expanded the intelligence of the calculator to accommodate the changes made to the Client Access Server and Transport components. This is all very good stuff that should be tremendously useful to those who have to lay out the hardware underpinnings of an Exchange 2013 deployment.
The default set of names chosen to name servers (as shown above, you can input your own names) are interesting as they include a German classical scholar (Mommsen), a French economist (Passy), a world chess champion (Fischer), units of radioactivity (Curies), a Dutch organic chemist (Vanthoff) and Ross, who needs no introduction within the Exchange world (clue for the unwitting: this is not Ross from “Friends”).
The standard caveats apply. Treat the output of even the most intelligent software tool as inferior to your human intelligence. No software tool can ever gather the kind of knowledge that you have of your messaging environment, its users, the business context, and the history of Exchange at your company. These factors influence Exchange deployments more than you might think and no software tool will ever be able to include all of the issues that need to be considered before any decision is made about hardware. For instance, this calculator makes no attempt to factor in the use of third party technology such as compliance software and you can bet your bottom dollar that this will affect Exchange configurations.
Microsoft warns that the output from this calculator cannot be compared to the output from the Exchange 2010 calculator. I think this is fair. The previous calculator makes certain assumptions and is based on knowledge that is no longer fully accurate in today’s environment. We know that Exchange 2013 is a different beast; it should therefore come as no surprise that its planning tools might produce different results to something based on Exchange 2010.
Of course, you could simply ignore planning tools altogether and buy the biggest honking servers that you can find, slap as much memory as you can inside the brutes, and connect them to a bank of the fastest disks you can find. You might have fun doing this and your local hardware salesperson will thank you for the generous commission that would result, but it’s probably not a good way to achieve management buy-in for your recommendations. So use the calculator and blame Microsoft.
Follow Tony @12Knocksinna