MSN Messenger services were down for over a week starting 3 Jul 2001. This is the service that Microsoft was touting as the basis of its forthcoming HailStorm. Can they survive?

I think that Microsoft understands what a huge black eye this is for HailStorm — people may forget in six months, but right now it is a big black eye on this beautiful picture they've painted for their partners. Surely American Express will ask some hard questions of Microsoft about what their contingency plans are for HailStorm. I know I would.

So, I think it is inevitable that Microsoft will have to do some major structural re-design. If you want to support the world on a 24-7 basis (and that is their goal, isn't it?), you have to be able to seemlessly handle billions of transactions per day if not per hour.

Now, I'm no Database Benchmarking guru — far from it — and I know that Benchmarks are the fourth kind of lie after Statistics.

That said, I'm thinking that the database to support HailStorm will have to be huge. And, I think it is instructive to look at the TPC-H benchmark where results are sorted by database size.

For the 100GB size, it is almost all about MS-SQL. At 300GB, they begin to fade, and by 1000GB there is no sign of MS-SQL.

Now, Microsoft can backend HailStorm with a non-Microsoft database and maybe even non-Intel hardware. HotMail, at least, used to have a UNIX backend. Perhaps this is why they have the core .NET runtime on FreeBSD?

But, they like to talk about eating their own dogfood, and I'm guessing that for such a major marketing push, running the core HailStorm database and services on non-Microsoft software would hurt too much.

Still it is important to remember that Microsoft has always sold software, not services. They've been content to have inexpensive but non-scalable software that they sell to other people. With the possible exception of MSN (and does MSN make money?), they haven't had to worry about making a profit from selling a service to other people.

As we've all seen, they can turn on a dime — at least when it comes to software and they don't have to change their business model: we give you software and you give us money and that is the end of our relationship unless you want to pay some hefty fees for software support.

The last major turn-on-a-dime trick was embracing the Internet — that is, refocusing their software development on the Internet. This time around it's different. This time around, they are going to try to get you to pay every month or year to continue to use their software.

And some users hear it and scream "But I already gave you my money!"

But there are two sides to this coin. If Microsoft doesn't provide the user with a reasonable service, the user doesn't have to pay and others will jump in and (to use Craig Burton's word) "redirect" that subscription.

After all, if you were a business owner who depended on Instant Messaging to do business and it's failing now, are you going to pay for this same level of service later when it comes to HailStorm? Seven days is a long time — a very long time — to be in the dark when you run a small business.


Microsoft doesn't really understand scalability or security. We've seen it with the numerous virii propagated by Outlook. We saw it with the DNS outage in January when disappeared. We'll see it again when they let any user (not just administrator level like UNIX does) use raw sockets in Windows XP. We are seeing it today with Messenger services being out.

To me, this also explains why those of us on UNIX-like platforms want to duplicate .NET and HailStorm. We tend to be proud intellectual snobs and Microsoft just hasn't gotten their act together after 25 or so years. They're still promoting BASIC, for goodness sake! Yeah, they got a lot of flash and bluster, but so what? Intellectual snobs are used to being ignored — we still know UNIX is better than Windows just the way that Bach is better than Brittney Spears. Millions of fans are always wrong.

It's axiomatic!