CORBA is a doomed technology, Ovum report says

The Common Object Request Broker Architecture (CORBA) is a 'doomed' technology that has no hope of matching Microsoft's Distributed Common Object Model, according to a new report on middleware. The report, 'Ovum Evaluates: Object Request Brokers,' was published by London-based research company Ovum Ltd. It also says that the biggest competition for DCOM will come from other middleware approaches, like message-oriented middleware such as IBM's MQSeries, or from distributed transaction processing monitors, such as BEA Systems' Tuxedo.

The Common Object Request Broker Architecture (CORBA) is a "doomed" technology that has no hope of matching Microsoft's Distributed Common Object Model, according to a new report on middleware.

The report, "Ovum Evaluates: Object Request Brokers," was published by London-based research company Ovum Ltd. It also says that the biggest competition for DCOM will come from other middleware approaches, like message-oriented middleware such as IBM's MQSeries, or from distributed transaction processing monitors, such as BEA Systems Inc.'s Tuxedo.

"The suppliers of distributed transaction processing monitors (DTPMs) are beginning to support object technology, which gives users the best of both worlds," says Rosemary Rock-Evans, the report's author. "You get the reliability, availability, wide applicability and security of a DTPM with the ability to use objects. And there is no need to use an ORB."

She says she favors this approach because most of the services are hidden from the user. "The problem with the CORBA ORBs is that you have to invoke every single service yourself, using a lot of commands. Some ORBs have 700 commands -- it makes development very difficult when you have 700 commands to learn," she says.

Rock-Evans says the battle between CORBA is reminiscent of the war between Unix and Windows, where the open standard is weakened by too many competing flavors. "None of the CORBA-based ORBs work together and the standard has no future," she says.

Compatibility is only part of the problem, however. The report takes up the weaknesses and inefficiencies of the CORBA architecture.

"CORBA is not really a very good standard -- there are lots of things wrong with it. The more I looked at it, I asked myself why anyone would come up with a solution like that," Rock-Evans says.

One example of this is in the CORBA object directory, which has to sit on a single node, and which does not allow object addresses to be replicated. This approach, she says, creates a single point of failure, and also slows performance because every user on the system has to access that node, however remote.

Rock-Evans praised the directory solution proposed by Microsoft. Called Active Directory, it creates a single directory that can be replicated around a network. This will take over from the current Microsoft Registry approach, which distributes a subset of information to various nodes, but which is complex and difficult to administer in a large system.

"Active Directory is a vital part of DCOM but it is a completely new approach and is obviously taking a long time to develop," says Rock-Evans. Active Directory has been delayed several times and is unlikely to be fully available before 1999, she says.

Despite the efforts of the Object Management Group to promote the CORBA standard, Rock-Evans says that very few companies are really using CORBA to develop large systems.

"Iona Technologies Inc. has the biggest market share [of the CORBA market] with around 2,000 customers, each with an average of five licenses, including client licenses. That does not suggest they are using it on a large scale. It suggests they are using it for experimental reasons, and could easily abandon CORBA if they wanted to," Rock-Evans says.

The report says that in the ORB market, the main winners will be Microsoft followed by Iona and Visigenic. But strong competition will come from existing middleware vendors that are adding object-based interfaces to their products, notably BEA with Tuxedo and NCR Corp. with Top End.

It pinpoints the main losers in the ORB race as Sun Microsystems Inc. and IBM and gives two main reasons for this.

First, both rely heavily on the Internet Inter-ORB Protocol, a technology that does not work out-of-the-box. They also have a strategy based on inter-operability -- something the other CORBA members do not strongly support and which may prove impossible to achieve anyway, as more proprietary services are added to the competitor products.

Secondly, both IBM and Sun favor the embedding of CORBA services in the operating system, which then compromises platform independence.

"Only Microsoft can get away with this sort of action, neither Sun nor IBM has the market share to make it work," the report says.

The report is particularly critical in its treatment of Sun, saying the company has a very weak position on the middleware market.

Ovum, in London, can be reached on the World Wide Web at http://www.ovum.com/.

Join the newsletter!

Error: Please check your email address.
Show Comments

Market Place

[]