By Unix News, a sister publication
There is no doubt that object technology will play a crucial role in computing. It is a significant paradigm shift, which the industry and users alike will one day accept as de facto. In particular, distributed objects have been touted over the last 12 months as a genuine solution to the growing problems with client/server architectures, as well as the way forward for the World Wide Web. As far as the Web is concerned, objects are indeed starting to claim the limelight. Unfortunately, the big object technology impetus is coming to life at a time when the IT community, across the board, has been plunged into a state of severe doubt over what to do with current infrastructures.
By Dom Pancucci
Faced with burning issues which range from client/server deployment to legacy integration, groupware, heterogeneous networks, intranets and the rest, the user class of 1996 is naturally a bit cautious about buying a ticket to another brave new world. Although objects have been around for years, it is only now that products based on the technology have started to appear in a distributed form.
Objective uses
Objects will deliver most benefits from operating in a network-centric model. Mainstream user organizations are concerned with what to do with current technology, let alone embrace something still seen as radical. And there is no doubt that moving to reliance on distributed objects is a big leap, however progressive the end results may be. It would be grossly unfair to say that the OT bandwagon has ground to a halt, but it has certainly reined back a little. The predicted rise of distributed objects has not yet occurred. There are good reasons for this. On the supply side, OT has gone through some big changes. The interest caused by
major developments last year have been skewed by events in the first half of 1996. One keynote over the last six months or so is that the Object Management Group has achieved widespread support for Corba 2.0, the standard to define a common request broker for object-based systems. Corba will be the foundation of distributed object systems, in particular where interoperability is important, so broad compliance will be essential. The launch by SunSoft of its NEO object-based version of Solaris was another 1995 milestone. NEO has joined NeXTStep from NeXT Software from which it has incorporated key technologies as a full function operating system based on OT. But headlines have been stolen latterly by two main developments. There is also a third important area where objects are becoming crucial intelligent agents. The most significant crowd pleaser in objects for 1996 is the rise and rise of Java, Sun’s object-oriented development language, now
under the wing of the JavaSoft business. The fact that Java is getting so much airplay is a compliment to Sun’s timing and perception of where the market is really leaning. This has to be set against the relative absence of Java-based products, although this will rapidly change. There are also problems reported with Java’s GUI, which has been found not to be entirely robust. Security is another concern with the development environment. The JavaSoft Web site contains a substantial list of
security related bugs and shortfalls with Java, as well as progress reports on problem resolution. Despite working hard to get NEO into release form, Sun must have realized that the commercial evolution of OT might end up coming from another direction. Having battled against the marketing chutzpah of Microsoft Windows for so long, Sun probably realized that a robust OT business would
start off at the client level first, only becoming important for the backbone once complexity issues started to grow. The bulk of object-oriented development has historically been on Windows platforms anyway although Java has clearly stolen a march with its platform-independent status. The second major talking point in this field is the sea change at Microsoft. The company has reorganized its drive into fully fledged distributed OT under the banner of the ActiveX group of technologies. In addition to many other things, Microsoft has aligned itself much more closely with the rest of the industry in the OT field.
Microsoft falls into line
ActiveX includes OLE, Visual Basic, Visual C++ and COM, the component object model which Microsoft is promising will scale to work with networks of hundreds of millions of clients and millions of servers. COM is built upon the Distributed Computing Environment standard with remote procedure calls (DCE/RPC) from the Open Software Foundation. Significantly, Microsoft is also promising to ease the path for the development of gateways to non-OLE software, such as Corba 2.0. Microsoft has also joined the Java camp and has a developer kit under preparation, cunningly called Jakarta. Yet these moves by Microsoft cannot be seen entirely in isolation. This year it has fully embraced the Internet as well, after dangerously ignoring the phenomenon previously. The rebadging of OLE and kindred technologies under the ActiveX heading has also given Microsoft a bit more room to redevelop what was known before as Network OLE the unspecified and so far elusive distributed form of OLE. For a while longer at least, Microsoft can enjoy the luxury of knowing that the bulk of OT development today across the board takes place on its operating platforms. It is the emergence of the World Wide Web which has changed attitudes at Microsoft, however, with the threat that a browser type interface could become a serious challenge to its mass market dominance in client
software. That threat is clearly epitomised by Netscape and its Navigator browser and server software. There is nothing quite like the fresh pulse of a budding mass market to get the sap rising at Microsofts Redmond headquarters.
Reassessing the future
Research firm Ovum released a major report on distributed objects late last year, which set out bullish prospects for the technology. Now the authors of the report have been compelled to reassess the situation. Katy Ring, senior consultant at Ovum is now more circumspect over distributed objects and plays up the roles of the software client and the Internet as factors driving current interest in OT. The market for distributed objects is more apathetic than we expected, which 18 months ago looked more promising. Those vendors with ORBs have also back peddled on their stated strategies to bundle the request broker with their environments. And users really don’t want the extra functionality of objects with the suggestion of hidden costs, says Ring. Originally Ovum identified a triumvirate of companies which will succeed in distributed objects Microsoft, IBM and Sun. Interestingly, Ovum saw Sun being largely restricted to its stronghold in the technical computing market, but is now prepared to admit that the company has scored a technical lead in the commercial market with Java. This puts the company in a position to challenge Microsoft. Yet due to problems with Java discussed earlier, Suns focus on the more secure corporate Intranet market is a shrewd and significant move. Java has catapulted Suns profile and it is up to the vendor to capitalise on this advantage, says Ring. The issue with Microsoft, however, is that although it may be another year away from delivering on its OT promises, the market will wait as it has always tended to. This is a golden opportunity for Sun, and IBM as well, to get up some steam. IBM received a good assessment from Ovum in its report and stood a chance of dominance so long as it got its act together over objects. Yet the world is still waiting for IBM to deliver. Ring points to the closer ties between IBM and Iona which sells the top rated Orbix ORB as casting further doubts over IBMs desire to push through the DSOM project to completion. The IBM brand name is still strong among users for OT, according to Ring, but the company is taxing this loyalty factor to the limit through its delays and apparent incoherence over OT. There are also few signs that the long awaited DSOM environment from IBM will tie up with Lotus Notes. Ovum has also reappraised the fortunes of NeXT Software, the pioneer in full function object-oriented operating systems. Its WebObjects product for making Web sites more interactive and resourceful has proved successful, with NeXT claiming that $2.5 million of WebObjects software licenses have been shifted this year. User organisations for WebObjects include Disney, Fannie Mae, MCI and Reebok. Neither is NeXT standing still it has written a development tool for WebObjects called Tsunami, which should boost functionality further. Ring is also convinced that the various deals that NeXT has with Microsoft will still be the root of its future success. Ring is convinced that the markets apathy towards distributed objects in the backbone will change once the client side starts to rock and roll. A tie up between the Java-like component vision for the Web and the business logic of companies is inevitable. And while the current drive is towards a thin client based on Java applets, the concept of a next generation GUI is slowly coming together. Ring believes this new GUI will be object-driven, including 3D graphics, multimedia and virtual reality interfaces. Its prospects have taken a step forward, following the
finalisation of the VRML 2.0 markup language for the Web. In May, Iona also teamed up with Neuron Data (ND) to bundle Orbix with Elements 2.0 the ND component development toolset aimed directly at the Web, OLE and Java. Jean-Marie Chauvet, VP for technology at ND, rightly points out that the market for interlinking heterogeneous OT environments will only grow if corporations see the economic benefits from migrating to a new architecture. Unfortunately this means that the initial confusion surrounding middleware will increase as many vendors vie for the corporate buyers mind share, observes Chauvet. Over the coming years, this confusion will be clarified as a limited number of standards emerge, according to Chauvet, leading to a high value marketplace for tools to build the new generation of
distributed applications. Object technology does provide the flexibility to achieve interoperable services and transport to create heterogeneous environments. Chauvet reckons that an object-oriented description of services is required which applies Web pages and accommodates the differences between COM and CORBA models. On the transport layer the essential requirement is the ability to convey objects, as they encapsulate both data and objects, in a peer to peer relationship, according to Chauvet. In current implementations, for instance, Java objects cannot be sent from the client browser to the Web server, or from Web server to Web server for that matter, forcing the developer to rely on other ad-hoc mechanisms to link servers, he says. There has been so much hype and false expectation regarding OT that it is amazing in many respects that the industry has sustained its effort thus far. And as users go further into the re-engineering of client-server and legacy systems, distributed objects will eventually appear to many to be a compelling solution. No one is prepared, however, to put a date as to when such a situation will occur.
