Still Waiting For Enterprise Deliveries From Microsoft

By Stuart J. Johnston

eginning this week, Microsoft officials will once again deluge the computing world with proud proclamations that Windows NT is finally ready for the enterprise. Indeed, they have been claiming for months that they have already made it.

Whether Microsoft finally has the tools to enable corporate IT departments to deploy enterprisewide applications or not -- and that is still an open question for a majority of users -- the company is still merely at the starting gate.

We're speaking, of course, about the delivery of the first of Microsoft's Enterprise Edition products, Windows NT Serve r 4.0 Enterprise Edition. While that product supplies failover clustering, transaction processing, message-queuing middleware, and support for eight-processor symmetrical multiprocessing machines right out of the box, there is still a lot more that needs to be delivered.

For instance, the company has yet to deliver SQL Server 6.5 Enterprise Edition, a release that supports databases up to 1 terabyte in size, and also takes advantage of the system software's ability to failover to a second machine if the first one dies. Without an application that takes advantage of it, the fact that the basic system will failover doesn't provide a lot of benefit.

And don't forget about a pending release of Exchange Server that will also failover. There are, obviously, applications from other competitors, such as Oracle and Lotus, that will take advantage of failover capabilities. These are still in the works as well.

Additionally, just failing over is not the only thing required to support a corporation's core bu siness processing needs. Microsoft's Cluster Server technology won't even begin to support advanced scalability by handling more than two nodes for a year or more.

Meanwhile, a cohesive distributed computing architecture will be needed in order to coax corporate IT to move production applications off of tried-and-true large- and medium-scale systems where they now mostly reside. Microsoft executives will be providing an outline of that architecture this week at its Professional Developers Conference in San Diego -- with product deliveries to support that blueprint beginning next year.

Windows NT 5.0, the core of that distributed architecture, enters beta testing this week. That is a far cry from the "been there, done that" attitude that Microsoft officials try to communicate at every opportunity.

Once it does have the majority of the pieces in place, however, Microsoft still faces the acid test that all enterprise vendors have faced over the years. That is, it has to prove to each and every custom er that it can do the job -- one customer and one application at a time.

That process is only begun at showy press events and keynote speeches at major trade shows. It is accomplished the hard way -- by designing, implementing, and supporting each individual application and making dead certain that it is not only solid 99% of the time, but solid 100% of the time. The applications that truly run corporate businesses do not have the option of going down for one hour out of every hundred. Can you imagine if the system processing your bank's credit-card transactions went down for an hour every four days? Or the system that handles your airline reservations? Or the air traffic control system?

In fact, I remember seeing a motivational poster at a corporate IT department years ago that said that 99% service would mean, among other things, a devastating airline crash at O'Hare Airport once an hour every day of the year. Thankfully, that is not the kind of service even our aging air traffic control system prov ides.

If you ask Microsoft executives, of course, they will say that they do understand that. But if you ask customers, many say they are not sure that Microsoft can deliver that, whether they claim to understand the concept or not. Part of that credibility problem arises from claims that Microsoft made in the past about its product offerings -- claims that it has sometimes been forced to acknowledge after-the-fact were not strictly accurate.

Call them gun-shy if you want. But corporate IT directors have learned the hard way with vendor after vendor over the years that they have to be "from Missouri" (e.g. "show me") in response to any and all claims of omnipotence, no matter who is making the claims.

So, even in today's superheated business environment, enterprisewide projects tend to take a long time to get going, because the applications are expected to last for five years or more once they are implemented.

One example of this, discussed in this week's IW 500 issue, is a standardized set o f enterprise applications for the oil and gas industry that more than 20 companies have committed to adopting over the next few years. The applications -- known as IS-Oil -- are a set of core enterprise resource planning applications for the oil and gas industry built on top of SAP's R/3.

Now, while IS-Oil can be implemented on either Unix or NT, by far most of the companies plan to implement it on Unix. That stands in stark contrast to SAP's claims that half of its new R/3 implementations -- not strictly IS-Oil, but all R/3-based applications -- are going in on NT.

What is happening here? The movement to create and implement IS-Oil started several years ago at a point when NT was still an unreleased product but Unix was catching on as an enterprise platform. Even today, some of the companies that are implementing IS-Oil on Unix -- but are using NT for other, less-demanding applications -- will tell you that NT isn't there yet in terms of scalability, reliability, and the quality of the infrastructur e tools necessary to support enterprise computing.

Microsoft has a pyramid-shaped chart that it likes to use to illustrate the price/performance advantages that NT has over Unix. At the very top of the pyramid is a little segment that constitutes a mere 5% of the total number of applications, but which accounts for half of all the money spent on enterprise computing.

That, according to Microsoft logic, illustrates that those Unix applications are too expensive and are vulnerable to their lower-cost NT onslaught.

Maybe. But there is an alternate way to look at that segment. That is where the cost is highest because those are the critical business-process applications that must do the job correctly all the time, never fail.

The fact is that corporate customers trust those applications and those platforms that they are already running because of their long experience with them. That trust comes because such applications follow a life cycle and many, if not most of those apps, are not yet nearing the end of their useful lives. Those applications work, so why "fix" them until absolutely necessary?

Last month, I had the opportunity to ask the CIO of one of the largest financial institutions in the world about the NT capabilities Microsoft demonstrated at its Scalability Day in New York last spring.

This CIO told me that processing a billion transactions a day was way beyond anything the company would need now or in the foreseeable future.

But asked if the company would consider implementing any of its enterprise applications on NT any time soon, the answer was "Maybe in four or five years." That would be 2002 or later.

Might the year 2000 problem speed up that process? The answer was a firm no. Why? "It would require us to rewrite every single line of code that we have," was the answer.

So while you will be hearing a lot from Microsoft in the next few years on how well it is doing in the enterprise, the process is bound to be much more tedious and painstaking than Redmond would like us to think. Indeed, many of the younger Microsoftians will undoubtedly be surprised at how slow and difficult attacking the bastions of enterprise computing is going to be.

It will require more than just hard selling. NT will have to prove itself in the most demanding environments, and that will take time. Microsoft has a long way to go before NT becomes ubiquitous at the highest levels of the enterprise.

It's a process that could give Bill Gates gray hair.

AuthorITies Archive

Send Us Your Feedback

Rich Levin: Run Time Rich fills you in on all of the latest products, issues, and trends in application development.		Stuart J. Johnston: Redmond Watch As our eyes and ears in Redmond, Stuart gives his perspective on the latest events at Microsoft.		Charles Pelton: Eye On IT Charles explores IT management issues and strategies that business and technology managers face.

Top of the Page