Through the Looking Glass of Microsoft's .NET
July 18, 2001
HailStorm, based on Microsoft's new .NET platform, is meant to be a giant database containing user-related information: credit-card numbers, calendars, contacts, e-mail, documents, and so on. Everything is going to be stored in a central location managed by Microsoft, if all goes according to plan.
Microsoft wants to make money by charging fees and subscriptions to access the HailStorm data. The company also wants to use HailStorm as the centerpiece of .NET, figuring that the data stored in HailStorm will "lock in" demand for lower-level .NET components, such as Windows XP.
The Messenger outage, however, reveals some thorns in the roses. For example, if the central location has technical problems, then people will not be able to get their data. A week without access to their calendar and e-mail would be a huge disruption for most people.
But I see a few, more basic problems with Microsoft's plan: People won't pay for HailStorm, and the lock-in won't work.
Some argue the "convenience" of having the data not tied down to a single machine, and therefore accessible from anywhere, makes it worth it. But I think the need to be "accessible from anywhere" is vastly overstated. Moreover, I can't see paying money for central storage, when I can get better reliability for free by keeping the data where it already is, on my home machine.
Even more interesting is the lock-in plan.
Microsoft has benefited enormously from lock-in plans in the past: the large numbers of users of the Windows platform has caused third-party developers to write large numbers of applications, feeding more users, in an ever-increasing spiral. Now Microsoft wants to do the same with the .NET platform, using HailStorm as the bait.
It worked well for Windows because a Windows application was either heavily customized to work on Windows, or ran on top of a middleware layer with the associated performance and functionality issues. Therefore porting a Windows application to another operating system was non-trivial.
The problem is, HailStorm is a platform that is accessed using standard network protocols (XML, SOAP, UDDI, etc). Therefore, any client that talks those protocols is equal to a Microsoft client built using .NET.
Imagine if Linux, Mac and OS/2 applications ran seamlessly on a Windows machine, side-by-side with Windows applications, with no performance penalty, and you have an idea of the difficulty of doing lock-in over the network.
Look at Microsoft's fight against Novell. Microsoft beating Novell in the network server market is the only real case where it has overtaken a competitor purely on product merits, without some "external" help.
For example, Word and Excel beating WordPerfect and 1-2-3 was helped by the simultaneous transition from DOS to Windows. (I say "external" meaning it was external to the products in question: of course, the DOS to Windows transition was also a Microsoft operation).
The reason Microsoft could overtake Novell is because a network server is essentially just a box on the network, and if another box talks the right network protocols, they are interchangeable. That is why companies like Network Appliance and EMC can make so much headway with network storage devices.
Consider Apache's lead over IIS in the Web server market, compared to Linux's slow attack on Windows. To replace a Linux machine with Windows, you need to worry about your hardware and applications that are running on Windows. Will they still be supported? Do you need to upgrade or replace them?
But in most cases, an IIS server can be replaced with an Apache server simply by plopping it onto your network.
Imagine a sales person trying to replace an SQL Server installation with Oracle (Nasdaq: ORCL) . Replacing them all at once is a huge risk and requires all the porting work to be done upfront, and replacing one machine at a time means almost as much work -- plus the hassle of managing two types of machines.
But if SQL Server morphs into the "XML-based .NET object database" that acts like a black box, communicating with the outside world only through standard protocols, then it is easy for a sales person to come in and replace one machine at a time.
So no matter how compelling Microsoft makes the HailStorm platform, there is no guarantee that this will automatically lead to sales of .NET plumbing.
You can't lock-in from the top down -- using HailStorm to generate .NET sales. Microsoft needs to do what it has done in the past: first get the platform -- .NET in this case, Windows in the old days -- established, then take advantage of that environment to create demand for applications such as HailStorm.
Adam Barr worked at Microsoft for over 10 years before leaving in April 2000. His book about his time there, "Proudly Serving My Corporate Masters," was published in December 2000. He lives in Redmond, Washington.