Original URL: http://www.theregister.co.uk/2005/11/09/ms2005_launch/

Microsoft takes on enterprise - again

Norfolk and Banks watch from the safety of London

By David Norfolk

Posted in Developer, 9th November 2005 15:14 GMT

I'm glad that it was our San Francisco correspondent, Gavin Clarke, who got to hear Steve Ballmer in person, challenging his competitors to have any doubts as to whether Microsoft's SQL Server 2005, Visual Studio 2005 (and the latest Beta of Biztalk Server 2006) are "enterprise ready". Ballmer is one of the few IT execs who physically frightens me. However, at the London launch of Microsoft's latest attack on Enterprise Computing -computing, that is, for large multinational companies doing large volumes of business- it was clear that this time it really does mean it. Perhaps it's personal.

Microsoft's prize customer Tesco.com commented that SQL Server 2005 was so much faster than its previous database that it could now move its model of the online shopping experience from too slow to feasible!

This is a notable victory for SQL Server 2005 (Tesco was so impressed it even got permission to use it in Beta), marred only by the fact, elucidated by a question from my co-author Martin Banks, that its previous database was the previous version of SQL Server. So the launch message is really that Microsoft customers, at least, had better upgrade.

No matter, as Microsoft is also targeting hobbyists and the midrange, so it's hedging its bets a bit - it's almost certain that SQL Server 2005 will be a big success.

There are "entry level" versions of SQL Server, eg., which are free downloads for the time being. This may well tempt some of MySQL's customers away from open source, although others will no doubt have religious objections to such a move. Microsoft already sells more database copies for mission critical applications, whatever that means exactly, than Oracle and IBM combined according to IDC, although IBM still gets more revenue from database sales than Microsoft does.

This probably means that enterprise customers believe that DB2 can do something that SQL Server can't and are prepared to pay for whatever this is (and remember that there's more to a database than performance on artificial benchmarks; management tools, third party support, security, resilience etc. are all important to the enterprise).

However, I have a sneaking suspicion that SQL Server 2005 might actually play rather well in the enterprise space, if enterprise users can overlook the inevitable hype. Dell, for example, was boasting of its 99 cent transactions in benchmark tests - but, in an enterprise context, what does that mean, unless you know what else they compromised to keep the costs down and how much real work its 99 cent transactions can accomplish?

Nevertheless, even if SQL Server 2005 can cope with enterprise computing demands, enterprise take up could still be slow. It takes time to convince conservative DBAs that a new database can be trusted; once they are convinced, it takes time to migrate large numbers of users; it takes even longer to regression-test the migrated business services and Oracle and IBM are improving their offerings too (although Bill Miller, general manager, mainframe business unit, of BMC, which already supports SQL Server 2005 with BMC SQL-BackTrack for Microsoft SQL Server and BMC Performance Manager for databases, claims that customer take up of DB2 v8 is slower than IBM hoped, probably for similar reasons).

At which point Martin Banks adds, "Never one to miss a blindingly obvious 'cheap shot' when it's presented on a plate, I did note that, although billed as a database ideal for user migration, all the users brought out for the launch were existing SQL Server users, so it was hardly a stunning demo of migration in action.

"To be fair, Andrew Lees, Microsoft VP of Server and Tools Marketing, did point out to me that, as the database was only just being formally introduced, it wasn't surprising that these early users were existing SQL Server customers. Nevertheless, less fairly, the thing has been out on beta for some time, so a release-candidate migration example might well have been possible. Lees did, however, hint that (while talking pointedly about not making any of those forward looking statements that get US lawyers in such a tizzy) he expected to have some migration stories to talk about within six months or so".

The other finished product launched, Visual Studio 2005, is attacking the enterprise market too, as with its Visual Studio 2005 Team System tools it has now more-or-less caught up with the sort of lifecycle development support tools other vendors (such as Borland, IBM/Rational or Telelogic, to mention just three) have had for years. Microsoft, of course, claims to have been using its new tools internally for a long time, and to have just "productised" them for sale with Visual Studio.

Once again, at the launch the customers weren't completely representative: although representing big companies, they mostly came from small developer teams and weren't that interested in the Team System parts of Visual Studio targeted at large, distributed teams.

Again, enterprise users will probably want to wait for this development environment to mature a little before considering it seriously and may be concerned about the lack of full UML 2 support. After all, UML 2 is a mature vendor-independent OMG standard and, while smaller players may appreciate "domain specific languages" and simplified UML, enterprise developers can probably cope with, and perhaps need, UML 2.

The proof of the Team System environment will probably come with third party support and this is looking good - for example, the Conchango consultancy has its SCRUM project management methodology plug-in for Team System in beta; Borland is delivering an "analyst role" for Team System, based on its CaliberRM requirements management tool (requirements management isn't supported much by Microsoft); and a less well-known vendor called Tangible Architect is even working on implementing what seems to be a version of MDA, the OMG's Model Driven Architecture, for .NET, which uses Team System where appropriate. If enterprise developers need UML 2 but otherwise want Team System, someone will supply them with UML 2.

Even so, enterprise developers are usually fairly conservative because they realise that change can hinder business as well as facilitate it - there is always a risk associated with change, as well as a risk associated with staying put. Which means that we'll have to wait and see how popular Team System proves to be in competition with the emerging Eclipse-based lifecycle development environments.

Borland, for example, has its Together tools for Eclipse and, according to Anthony Kesterton, IBM Rational Technical Consultant, IBM is using its RUP expertise to help Eclipse go up a level and become less code-centric, initially in the Eclipse Process Framework project.

The most interesting part of this Microsoft launch event was the "chalk and talk" session, where Microsoft put up its real technicians. Some of these had comparatively recent experience outside of Microsoft and obviously did believe that SQL Server, say, could now go head to head with the big boys (although one of them commented that mainframe DB2 was a much tougher target than Oracle, or MySQL, itself a sign that he knew what he was were talking about). More to the point, they sounded believable when they talked of successfully migrating mainframe database installations to SQL Server, although they couldn't talk about these in detail yet.

Martin Banks, however, has his doubts: "I actually don't see migration as a major or important target for this or any other enterprise tool anymore. It is one of the potential advantages of service orientation that users will be building loosely-coupled, composite applications that integrate legacy databases - or the bits of them that are relevant - rather than being forced into a decision that brings the dangers of database 'rip and replace'.

"Old databases will be around as long as the services they provide are valued more highly than the cost of the licence. But once that balance goes negative, and other cheaper databases can do the work well enough, old databases will be out on their ear. So it'll be migration by osmosis rather than radical surgery".

So, in conclusion, an interesting launch and one that does point towards Microsoft really making it as an Enterprise player at last - at least once Service Pack 1 has arrived. However, we'll have to wait to talk to, say, Oracle customers who are actually migrating to SQL Server, or adopting it for new business services, before we can see how realistic Microsoft's Enterprise aspirations really are. ®

David Norfolk is the author of IT Governance, published by Thorogood. More details here.