Life may not get any quieter for hard-pressed corporate IT managers in the aftermath of Y2K. Vital decisions will have to be made as pressure grows to migrate IT systems to the all-important n-tier model and to make them Internet friendly.
As one industry expert puts it: "Choose right and their companies get to compete in the lucrative world of Internet commerce. Choose wrong, and they will be serving lunch to their competitors".
"Host to web" (i.e. running applications over the Internet as opposed to each computer having an installed copy of the software) and n-tier projects head the list of post-Y2K priorities.
"N-tier systems are inevitable" according to Paul Hickey, vice president of Product Alliances at Iona Technologies, which markets n-tier enabling software. Predictions by the Gartner Group place a $3 billion value on the potential market arising from these global strategic shifts by 2002.
But what is the n-tier model? It's a view of applications as being composed of three elements which constitute the fundamental "tiers" of n-tier systems. These are: user presentation (or Graphical User Interface), business logic and data services.
Such systems have the advantage that their composite tiers can be individually distributed over a network. So enlarging a company's IT system, for example in response to greater sales, is made easier.
Companies which had previously followed client/server-based IT models generally lacked this extensibility and often found themselves locked into inflexible arrangements with proprietary system vendors.
The Internet is proving to be the ideal "delivery mechanism" for the systems based on the new model. "Several years ago, it was often a struggle to explain to people why they should move to an n-tier architecture, but with the opening of a lot of their internal systems to the Internet, scaleability became critical and the only way you get that is by the n-tier architecture," says Saul Cunningham, a systems engineer at BEA Systems's Sydney headquarters.
Many companies will seize the opportunity to migrate their IT systems to the n-tier architecture and the Internet in one operation.
Roger Sessions, President of ObjectWatch Inc., which runs an online open discussion forum promoting the advancement of n-tier technology, anticipates that "the move to Internet and ntier will happen concurrently. The Internet pretty much requires a 3-tier architecture to support the scaleability requirements".
The expected post-Y2K demand will spur growth in the already burgeoning middleware and application server markets. It seems certain that it will be one of the industry's most buoyant and hotly-contested business sectors.
Middleware vendors, including Iona Technologies, BEA Systems and Inprise, are weighing in with major product announcements this year to capitalise on the corporate migration to the "next generation strategic architecture" as it has been billed.
The term "middleware" refers to the set of software tools which facilitate building and running of n-tier applications. Companies could opt to forego use of this type of third-party software and develop in-house n-tier infrastructures but, for many, the costs and technical complexities are prohibitive.
With the pending CORBA 3.0 release later this year, there is widespread speculation that the CORBA and EJB specifications will effectively merge, possibly giving rise to yet another chapter in the "Microsoft versus the rest" industry saga.
Companies which successfully migrate core business applications to the n-tier architecture can contemplate extending the framework to encompass all its in-house developed software products.
In simplified terms, this is done by using middleware tools to transform the existing software used by each company division into component form. Such systems, though large, would be modular in construction and thus easier to administer and maintain.
But the big question facing many companies contemplating embracing the ntier architectural model is which architecture and vendor software to choose. It is a decision of enormous importance which may affect their business prospects for decades to come.
Commercially available middleware products follow one of three competing architectural specifications:
The Microsoft candidate, COM+, will be a key feature of its next version of Windows NT (to be known as Windows 2000).
Sun Microsystems's offering of EJB (Enterprise Java Beans).
CORBA, which has evolved over eight years under the aegis of the OMG (Object Management Group) alliance of companies.
To some experts, it appears the field looks set to narrow to a straight Microsoft's COM+ versus Sun's EJB head to head, playing down the significance of CORBA-based packages.
That notion is strongly rejected by Paul Hickey: "Today CORBA is the only distributed n-tier architecture that's been proven and tested and has been deployed in any kind of significant levels". He considers it instructive that a leading telecommunications carrier in Europe records up to 50 million CORBA invocations a day.
COM+ gives users the benefit of Microsoft's seven-year evolution of this system and its guaranteed usability on an operating system platform which dominates over 90 per cent of the domestic and office software markets. But questions over the scaleability of Windows NT/2000 may cause concern.
EJB benefits from the Java language slipstream which has been exciting the industry for three years now. It is claimed to be independent of operating system and middleware and to interface with CORBA with ease.
Technical comparisons between EJB and COM+ have also focused on the algorithms employed by both in the management of large numbers of client requests to servers.
CORBA, on the other hand, has strong portability claims and over time has proved successful in breathing new life into costly software which would otherwise have been abandoned. CORBA 3.0, due for finalisation later this year, is set to make the specification easier to use and to tighten its integration with Java.
liamod@aol.com