Since its founding in 1989, the Object Management Group (OMG), Framingham, Mass., has pushed to create standard specifications for object-oriented software and component-based development. The group is funded by 700 software and computer vendors and user organizations. The OMG Common Object Request Broker Architecture (Corba) specifications are finally gaining widespread acceptance as the group vows it will complement Microsoft's Distributed COM desktop alternative. Here, OMG President and Chief Executive Christopher Stone discusses with Managing Editor Michael W. Bucken the evolution of the consortium, its newfound prominence in the enterprise and Internet development spaces, and how the group continues to try and outdo Microsoft in the battle to create object standards.
Corporations and software vendors were slow to accept Corba in the early years of OMG, but that seems to have changed over the past six months or so. What is the key reason for the change?
I think a couple of things helped us in the last few months. I think companies were waiting to see what happened with distributed objects. Was Corba to be the next infrastructure to move foward or was Microsoft going to do something? I think they finally decided that from a scalability perspective, and for a back-end development platform, Corba makes the most sense. The time was right for us.
We still have the early people, but what I've seen over the past couple of years is that large software companies are using the OMG specs as sort of the basis of their infrastructure. We've recently seen Netscape One using Corba and IIOP (OMG's Corba-based Internet Inter-ORB Protocol). Oracle's doing the same thing with WebServer.
Have corporate I/S organizations started to widely embrace OMG standards?
We've seen a fairly radical take on the Corba specs in the user community lately, meaning in the last six to eight months. The same issues had held it back there -- the issues of: 'What's Microsoft going to do?'; 'What's Netscape going to do?'; and 'What's the Internet going to look like?' A lot of people were scared. But finally they are making a move.
Is Corba finally ready for enterprisewide development?
It looks like Corba is there. It's not going to go away. It's had it's two to three year maturing phase. It took longer than expected, but that's software. I always thought that 1997 was a breakout year, and it still looks like that might just be right. We're predicting shipments of Corba-based servers and runtime licenses will be in the tens of millions this year mostly thanks to Netscape.
Do you still have to worry that Microsoft will successfully create an object standard that can supersede Corba?
Yeah, sure. The issue here is that there are two different design centers. Corba was developed from the network and scaled from the network to begin with. DCOM is an extension of the desktop. DCOM doesn't even use TCP/IP. The thinking is kind of odd, but it's desktop thinking. I think people now have come to the conclusion that there has to be some kind of bridge or interoperation between the two. You will see a lot of DCOM and COM in desktops and small server configurations. And you'll see Corba and IIOP in very large Internet apps and in a lot of back-end systems.
"I always thought
1997 was a
and it looks
like that just
might be right."
Are you concerned that The Open Group has Microsoft's blessing to create
an ActiveX standard?
Remember, we only do specs. And they do code. So Microsoft really had no choice
of where to go. The Active Group, which is the organization that Microsoft has
put together to look directly into it, is part of The Open Group. The Active
Group is still in its infancy. They still haven't really done much so the jury's
What is OMG's relationship with Microsoft?
We're trying to cooperate. I believe that you will see cooperation between us, directly, in the next few months because I think we're both kind of tired of reading about this friction. I think we're finally coming to respect each others' abilities and installed base. We just have to work with each other. One is not going to supplant the other.
Vendors of message-oriented middleware products maintain these systems are complementary to Corba ORBs. How does the OMG position Corba versus message-oriented middleware?
We don't compete. As far as we're concerned, asynchronous messaging is a useful transport mechanism for objects. We're adding asynchronous messaging onto Corba, within the next three months there will be a separate asynchronous messaging model. That will be part of Corba. Does that mean I'm trying to bury the MOMA guys like PeerLogic? No. It just means that I firmly believe that distributed objects using a standard spec like Corba will be the middleware of choice. And any way that I can promulgate that, I'll do it.
When will you be choosing the asynchronous messaging technology spec, and whose technology is in the running?
We'll probably get 10 responses (to an RFP for the technology). We've already got responses from IBM, MessageQ, DEC's asynch series, DECmessageQ, Peer Logic, Seer Technologies and all the middleware players. The result will be a standardized spec on store and forward using Corba. We should have it done in six months.
What is the status of the effort to create a design standard that attracted proposals from Rational Software and Platinum Technology?
There's an interesting contest shaping up because Microsoft is responding to the RFP with Rational. That proposal includes the UML (unified modeling language developed by Rational). We didn't ask for the UML, we asked for interoperation on design models. Rational is giving us the UML. It remains to be seen whether we will actually adopt all of it or not. What we are looking for is interoperability between models so that you have a meta object facility. Platinum is in the other camp. There may be another proposal. We'll end up choosing one in the end. That will shake out in the next six to eight months as well. That will be a big win for us if we can get agreement on a meta object facility, a design model. That could potentially be a unified language for design.
How does the OMG see the future of the Internet and Internet development?
My position is that we (the OMG) will control the infrastructure of the Internet, period. That will come next year. There are three infrastructure options to continue as is, which is stateless HTTP. The second is DCOM. The third is Corba. Will you see a melding of the three? Will you see Corba IIOP and HTTP melding together? There is some activity going on to do a merger of HTTP and IIOP. But fundamentally, I think that by this time next year, most of the Internet apps -- the new apps -- will be developed using Corba IIOP.
Why will Corba IIOP win?
All the big players, IBM, Netscape, Oracle, are promoting the use of this in their development environments. Lotus' Domino, that's IIOP. Netscape's Galileo is based on IIOP. This time next year there will be 20 million runtime and server licenses of IIOP. In 1998, 1999 who knows, but that's a big number for next year. Netscape skews the number way out of proportion, with desktops and runtimes, but we've talked to them and those are the numbers.
Corba has been criticized by some analysts for not fully supporting Java. How do you respond to the criticism and what is the OMG doing to improve Corba support for Java?
Java and Corba already go together very, very nicely. You will see a lot more cooperation between Java and Corba as we support more services and as the virtual machine concept of Java becomes more rounded out with more services, like security service and transaction service. You'll see a lot of Corba integration. There are probably a dozen Java ORBs out there already.
What are the OMG's plans to enter new pursuits? Do you set limits as to where you'll go?
One area that we're thinking about is numbers. Actually getting into doing some research, providing statistics and analysis on the growth of the object market; market data basically. Do I want to compete with IDC and Forrester? Not really. I just want to focus on what we know, which is distributed objects. But I have been and want to continue to be extremely focused.
Mike Bucken is former Editor-in-Chief of Application Development Trends magazine.