The bully is back

Microsoft's server strategy is to make enemies suffer, fence-sitters choose sides, and customers pay up

AT A BLEAK TIME when other vendors are pulling in, laying off, and riding out the recession, Microsoft is partying like it's 1999. But not everybody is invited. If you're a Windows ISV (independent software vendor), you should make sure Microsoft hasn't written you out of the picture. If you're a Windows or Microsoft enterprise software competitor, it's time to shelve the strategy you've been using for the past two years to market against a humbled, rudderless Microsoft; that company no longer exists. If you're a current or potential Microsoft enterprise customer, you'd better brace for the hard sell in 2003 and a big spike in licensing costs.

By piecing together what Microsoft tells each of its constituencies, we can make some predictions about where Windows and the Windows server stack are going and how changes will affect customers and competitors.

Before 2002, Microsoft welcomed heterogeneous deployments, in part, because it knew its server software was not up to snuff. Windows 2000 was loaded with serious security and scalability problems, and the COM API was getting hammered by the cross-platform J2EE. Exchange Server had all but withered from inattention. SharePoint got little notice in the portal and collaboration markets. Commerce Server was eclipsed by IBM's WebSphere. ISA (Internet Security and Acceleration) Server failed to steal share from companies selling caching servers and firewalls. Microsoft's enterprise juggernaut had run aground.

Once Bill Gates got out of his way, Steve Ballmer and his most loyal officers quickly set about remaking Microsoft inside and out. Internal development and review procedures were completely overhauled to keep sloppy code from leaving the shop. Ballmer's "ass on the line" policy sets up a chain of accountability that runs from worker bee developers to vice presidents for every product. That took care of new software, but Microsoft also needed to polish up its existing product line. Ballmer's team figured out that Microsoft can't fight its enemies if it's wasting energy on infighting between groups. Turf squabbles were proscribed and everyone's focus was turned outward to real threats in the enterprise: Sun, IBM, and OSS (open-source software).

Microsoft's other enterprise foe is its own image. To turn this around, two parallel campaigns were launched. The technical mission revolves around security. Microsoft wants to shatter the presumption that Windows and its components are inherently vulnerable to attack. Programs such as "secure by default" and "secure by design" apply primarily to the Windows operating system, but secure coding paradigms are enforced throughout the product line.

The other campaign is to correct the company's infamous tendency to stay mum about its products' flaws. Microsoft's management, including Gates himself, has been dispatched to deliver a message: The remade company is willing to own up to its shortcomings and failures. Through candor, Microsoft hopes to engender the kind of forgiveness the public grants more sympathetic entities such as Apple and open source. However, that frankness does not necessarily extend to the accurate positioning of its enterprise software in the overall market.

Shell game

The enterprise server stack will migrate to managed (.Net-"safe") code slowly, initially taking on .Net interfaces and scripting capabilities. The reason? Microsoft has millions of lines of existing code; porting wholesale to .Net would be too expensive. That's a frequent customer objection to .Net that Microsoft can't counter with its own experience.

The .Net interfaces to Microsoft's server stack will get traction, if only because they'll be the default for developers: .Net's standard messaging engine will be Exchange, its collaboration engine will be SharePoint, its database engine will be SQL Server, and so on. The new "better together" philosophy -- nothing but Microsoft and Microsoft-approved servers and clients throughout an organization -- will be exemplified in the .Net API and in Visual Studio .Net tools.

In licensing, Microsoft is already selling as though Sun is out of the market. Every server software component carries a separate per-seat license fee. If you use a Windows client to connect to a .Net Server machine running SharePoint and Exchange, that's three client licenses: The OS, SharePoint, and Exchange. If you also use a Web browser to view charts drawn by Project Server, you need a $179 client license just for that purpose. According to Microsoft, that license is per device, not per connection, so you can't pool licenses.

If you think you can cleverly sidestep this scheme by running Linux or Unix desktops or gateways, forget it. Anything that invokes, directly or indirectly, a piece of Windows server functionality is subject to a client license. That extends to non-Windows clients. Let's say you put up a Linux file server and run the Samba Windows file service. When a user attaches his desktop to a Samba share, Samba reaches out to your Windows Active Directory server to authenticate the user. The user is pulling files from a Linux box, but authenticating through Active Directory. That makes the user a Windows client.

With companies rushing to cross-connect their server, desktop, and mobile assets, it's hard to imagine how a company with just one Windows server could avoid buying Windows OS and enterprise client licenses for every network-enabled device.

Microsoft's standing policy of leaving room for ISVs is being reevaluated, case by case. For example, old Microsoft pals Symantec, Executive Software, and Citrix will find their markets for PcAnywhere, DiskKeeper, and the Citrix multi-user Windows platform decimated by standard features in .Net Server OS. Substantially beefed-up, virtualized networked storage capabilities target Sun but will hurt EMC, HP, and IBM as well. Microsoft's romance with Groove excludes other players in the peer collaboration space.

The vendor's dalliance with open-source software (for example, BSD and Linux editions of the .Net platform) is self-serving. Microsoft risks little by allowing OSS ports of .Net. OSS hackers aren't going to create open-source versions of Exchange or SQL Server. "Better together" is Microsoft's OSS defense; the combined Microsoft server stack is .Net's back end. Worthwhile .Net enterprise software will expect most of that stack to be present, just as worthwhile Java enterprise programs require J2EE. Besides, pushing .Net to other platforms only creates a broader base for licensing.

It may look like we're ragging on Microsoft for taking an aggressive position. On the contrary, we think the company is taking bold, shrewd steps that will consolidate the software market and raise the bar for enterprise servers. Some companies are so tired of hand-wiring discrete solutions together that they'll happily write a check to Microsoft to cut their integration hassles down to size. The question is, will most customers buy the whole Microsoft enchilada -- ranging from the awesome .Net Server OS and SQL Server database to the more mundane Exchange and the irrelevant ISA Server -- or will they insist on integrating Windows with their existing enterprise solutions? It's a crap shoot. But like wily co-superpower IBM, Microsoft is setting itself up so that no matter how the dice roll, it wins.

Copyright © 2002 IDG Communications, Inc.