Interview Later this year, Sun will start shipping almost every piece of software it makes from the Sun One Application Server and Directory Server to grid computing and clustering products as part of a giant bundle with its Solaris operating system. This Microsoft-like bundling exercise has the potential to change the way both software and hardware are priced. Early indications have this massive software suite shipping to customers at a cost of between $100 and $200 per employee, which would be quite a blow to the thousands of dollars per CPU middleware crowd.
Sun has tried for some time to boost its middleware business with modest success, but McNealy, Chairman, CEO and President at Sun, is convinced that this aggressive pricing model will make Sun's big software bet pay off. Sun plans to use the Orion stack as an attack against HP, which abandoned its own middleware line last year, and against IBM's WebSphere products.
Sun's rivals dismiss Orion as just another ploy by the server company to generate much needed software revenue and doubt it will work. McNealy, of course, is confident it will succeed.
When did you first come up with the idea for Orion? Did HP's decision to exit the middleware business help push you in that direction?
It wasn't as much a competitive thing as it was a customer value thing, as boring as that sounds. When we put all of the Orion software in Solaris, guess what happened. It didn't work. So we debugged all that and now it works.
But what is the competitive angle behind Orion?
We looked at WebSphere, and we are the number one platform for WebSphere, and they are siphoning off all this money. We have an awesome Web server that runs like the wind. It has availability and reliability features like you can't believe. You also don't have to go out and buy another app server. There might be a reason to go BEA on the very high end, but 90 percent of the users will be just as happy not spending money on that.
Why should all this stuff be siphoned off by these middleware folks, leaving you with users who don't have enough money to go out and buy computers and computer systems. It's really more of an attack on WebSphere than it is on HP, but if you buy an HP computer, you still have to buy the stack. You still have to go buy an app server, a Web server, a directory, calendaring and all the rest of it.
They are each pigeon-holed, probably HP more than IBM. HP has to go to the outside. IBM in the sense that they have this huge multi-billion dollar software business that is going to come under unbelievable pressure because all of sudden only their servers are the total available market and secondly, they have to get cost competitive.
What's the motivation for a customer to pick your software over the market share leaders?
This is where the CIOs typically don't do the right thing. They go, "I am going to do the total cost of acquisition of the server. Then I will do a total cost of acquisition for the app server and the directory." If they add them all up and look at the total acquisition cost of the system, we win hands down. We are so far and away the low cost producer. With us doing the R and D on the software integration, there is no integration costs for the customers, which never gets factored into the total analysis.
Our equipment is higher volume and less expensive than theirs, and there is no IBM Global Services needed to integrate it anymore. They can't charge $40,000 a CPU for just the app server.
But no matter how you slice it, IBM owns the database with DB2, which is the biggest piece of the software stack. They can leverage that against you in their pricing to a large degree.
That's why we partner so aggressively with Oracle.
So that still gives IBM more control than Sun with regard to the overall cost of software.
They have a business model and a revenue model that they have to go support, and that database revenue is pure profit. Where MY SQL works we say use MY SQL. Where the directory database works use the Sun One Directory Server. Where Oracle works use that. There are lots of other database strategies that are being developed out there.
It's MIRVing. Databases are MIRVing.
IBM and Oracle would rather have you believe that everything is going to go into one database. In fact, that is not what's happening. The world is MIRVing.
Multiple Independent Re-entry Vehicle. You didn't know what MIRVing was? You're not old enough to know.
During the Cold War, the issue around the arms race wasn't just how many missles you had but the fact that the warheads could MIRV, which meant they would go off and you could have seven independent, targeted pieces come flying off. So you could hit seven cities with one missle.
This goes back to a time when in elementary school we used to do air raids where you had to get under your chairs and cover up.
So with databases MIRVing, there's just data showing up in lots of different places. There's a database in StarOffice. There's a database in your phone and all of your appliances.
That still gives IBM a way to undercut Orion though?
I don't know that it is the dominant piece. It is a piece of the software stuff you have to pay. The real question is how many more database licenses do people really need. I don't think they need all that much more, and we don't have any revenue model built around that.
Yeah, I wish we had one. But we do have one - MYSQL, Postgres and the Sun ONE database technology.
I think the open source opportunities are non-trivial there.
Have you considered making a database acquisition?
You know, we haven't decided that is a war we want to go fight. Why not let them all beat each others' brains in.
It's still unclear to me how users will purchase Orion. You are giving them the whole stack whether or not they just want one piece. Do you expect them to move to Sun one piece at a time?
The stacks are always up for upgrades. That is the only way the stack-makers can make their next quarter numbers. At some point, the license expires or they need an upgrade. It is not like overnight everyone is going to go to Orion. At some point they say, why don't I just go because it's cheaper.
Do you have an idea of what this migration might look like?
We don't really know. It will be an interesting experiment.
I would buy an Orion license and then tell everybody that before we ever sign another software license tell me why we shouldn't use what we already own. The real goal is to price it aggressively enough so that if you are just going to use one piece of it, it's still compelling.
At that point, I thanked McNealy for the interview and tucked my computer away in an HP-branded backpack. It has lots of nice pockets, you see.
McNealy wished me good luck and then asked an associate, "Can you get him a new backpack?" ®