Wikibon Blog

I’ve been doing a bunch of research on cloud-based email lately. It started when I was speaking with a  number of Wikibon members about what they’re doing with Exchange and whether Exchange 2010 is changing the way they think about deploying storage. The Exchange team in Redmond has been pushing DAS harder than I’ve ever seen. Microsoft is telling customers that it has dramatically reduced the IO load from Exchange 2003, which is true.

 

Exchange 2003 was IO bound and very ‘bursty’ meaning performance was unpredictable. I believe the technical term for this is Exchange 2003 is a pig. So SAN has been the obvious choice for many Exchange deployments. Exchange 2007 addressed much of this IO problem but many clients skipped right over 2007, waiting for Exchange 2010. Wish I’d taken that path with Vista.

Microsoft is now pulling out all the stops. It’s telling clients that I spoke with to think about maintenance expiring on Exchange 2003 (I think you can still buy an extended service plan through 2014 if you give up your third child) and that SAN is not a recommended configuration for 2010. Microsoft is telling customers to worry about complexity and SAN can be a single point of failure. The logic put forth is that if I lose a DAS device I only lose part of my storage whereas if my SAN goes out…all my data is inaccessible. Interesting logic I thought.

I started to think about Microsoft and how they’ve behaved over the years and I thought that Microsoft must be trying to take back some of the value they’re losing to storage vendors and that’s why they’re recommending DAS so aggressively. It’s clear, right? They’re trying to put more storage function like replication into their own software stack and make their software more valuable to users…we’ve seen it a zillion times. 

Except…I spoke with Microsoft about this and they said that really wasn’t the motivation. In fact they said many inside Microsoft love SAN. Rather Microsoft indicated to me that the vast majority of problems with Exchange can be traced to storage so Microsoft figures they’ll do the industry a favor and improve certain Exchange functions and simplify the whole environment; which will make Exchange better and more reliable. Thanks Mr. Softy , that’s so nice of you.

Being the skeptic, I starting to think about this more deeply to try to find the real motivation behind these trends. Then it hit me in a moment of mastering the obvious; something I’ve grown quite good at.

Google. Duh! Everything these days goes back to Google. Of course. Here’s an oldie but goodie updated to show the next wave we’re about to see in the technology industry. 

 There are three key points I always make about this chart:

1. The waves keep getting bigger. 
2. With each new wave, some big companies die and some small companies get big. 
3. Each wave has a big daddy and this coming decade it’s Google.

No company has more to lose in this transition than Microsoft.

The Google Apps Tsunami

Now I’m not predicting the death of Microsoft mind you. I don’t see that happening. Too much cash, too smart, too many good people. But I am here to tell you that Google Apps is getting into Microsoft’s shorts big time, using plays that Microsoft has never seen before. This isn’t Novell or Lotus or Netscape or Sun or even Oracle. This is Google– the advertising monster. And it started with a blank piece of paper and it’s driving Microsoft nuts. Microsoft simply doesn’t have a good answer.

Check out this data that I and some of my colleagues put together comparing on-premise Exchange costs with Microsoft Exchange Online and Google Apps for a 10,000 seat installation.

Microsoft Exchange on-premise is about 4X more expensive on a TCO basis than Google Apps; and Microsoft’s SaaS offering, Exchange Online, is about 1.6X the cost of Google Apps, even though its monthly subscription is less (check out your pricing with this calculator from Microsoft; here’s Google’s pricing– no need for a fancy calculator). And the big pieces of the cost bar that Microsoft wants to lower are storage and archiving (which is really storage). Oh, did I mention that Microsoft doesn’t get any of this revenue today so it has nothing to lose by telling all of its customers to bail on SAN.

So the bottom line is Microsoft wants to bomb storage costs so it can make on-premise Exchange more competitive with Google and preserve its obscene margins from Exchange and client-side software. It might actually work for a while.

But is DAS really cheaper than SAN for Exchange? Well, it depends. In my next post we’ll look at a model we built to answer the question: Should you use DAS or SAN for Exchange?

Happy New Year people…the Surf’s up!