The CEO of one of the largest software companies in the world plans 50 Linux applications by year's end. Here's why.
So what does the CEO of the world’s fourth-largest software company think about an open source phenomenon like Linux? Well, for one thing, that the “open source” side of things is not the interesting part. What has Sanjay Kumar all fired up is the enthusiasm CA’s enterprise customers — mainframe users who wouldn’t give Linux a second glance two years ago — suddenly have for the platform. And Kumar should know what the mainframers are thinking — they make up about 50 percent of CA’s business. That enthusiasm (for Linux, not free software) has caught on at CA, which claims it will have over 50 applications ported to Linux six months from now. Linux Magazine’s Adam Goodman, JC Freed, and Robert McMillan caught up with Kumar in New York at the January 2002 LinuxWorld Expo to ask him about Linux in the enterprise, the GPL, and the $64,000 question: whether or not IBM should do its own distribution.
LINUX MAGAZINE: What did you first think of Linux?
SANJAY KUMAR: My first question was, can I make money from this? I mean, that’s what this is all about, right? And I had to be convinced. If you want me to say, “Sign up to invest in this thing,” somebody has got to convince me that the world is going to support this platform. And it took a few rounds to get there. It was as much an exercise for us, internally, to think about the platform and do some research on it.
I started talking to customers, because I believe in having a lot of independent views on things, and what I found at that time — 1996-1997 — was that high-end customers didn’t want anything to do with it. Low-end customers were incredibly excited because of the price — especially if someone was going to support it. And mid-tier customers were willing to buy bundled solutions. In other words: “I want a turnkey solution. I don’t care what it runs on, I just want it to work. I’m buying the application.”
And so in 1998, when we started doing more work on it, I expected the low end to be there, the midrange to be bundled with applications, and the high end to be non-existent. And I can tell you I was completely wrong. The high end is where the action is today. I mean, you have the core Linux community, right, that’s been spread around: either the mid tier, the low end and some of the high end. But in terms of who’s willing to spend money, where we thought we would start and where we are today is night and day.
LM: So what happened? What enabled that?
SK: A lot of CIOs were being talked to about Linux by technical people, but the pushback was, “But if it breaks, who do I call? If something doesn’t go right, you know, who do I blame? There’s no one person behind it.” And that’s very unnerving to people. There has not been a technology around that has been incredibly popular, that’s not had somebody whose feet you can hold to the fire and say, “You know what, if you don’t fix that problem, I’m not buying anything from you in the next quarter and you’re going to suffer.” That’s a powerful hammer. People didn’t have that [with Linux], and that’s a major cultural adjustment for customers, because they want accountability. And when IBM stepped into it last year and said, “This is kosher; we’ll stand behind it,” I think that brought a lot of people to the edge, to say, “OK I’m willing to really consider this now.” That was the big event.
At that point we authorized the cranking up of two development labs for Linux and porting a whole bunch of products. And we’re seeing the fruits of that labor today. We placed a bet that IBM was going to get behind it very seriously, and they did.
The Back Door
LM: IBM’s sponsorship of Linux may appear as a double-edged sword to you. On the one hand, they legitimize the platform for your customers, but on the other hand, do you feel that there’s a risk that IBM may somehow be overly associated with Linux …
SK: You mean hijack the platform?
LM: … If you want to call it that. But also, IBM seems willing to commoditize a lot of lines of business, so that could also present a problem for companies like yours.
SK: We have competed with IBM for 25 years. We started with one product. We sold it and they gave theirs away for free, so the idea of them bundling — while it’s always a formidable thing to compete with — doesn’t bother me because ultimately the customers are going to pay for value. If you’ve got differentiation and value, [customers] will buy. We’ve learned that the hard way over 25 years.
The one thing that’s counter to the argument about any one person hijacking the platform is, by definition, the platform’s portability. I think what will happen if customers start to deploy on [IBM's] Z platform for Linux and the hardware gets too expensive, people are going to say, “That’s fine. You know what? I’m going to invite Dell, Compaq, and HP to come in and bid a nice 16-way cluster server set to just move the platform over.” And a number of customers that we’re working with today — high-end household names — are using the Z platform to develop and test on, but will deploy on Intel.
LM: Why is that?
SK: Because they believe, from a pricing perspective on hardware, the Intel platform is a commodity and the Z platform is a monopoly. Think about that. There is only one choice for Z.
I can tell you there are some large customers who six months ago were committed to deploying on Z. I had a conversation with one of them, a super-large customer downtown in the city [New York], who told me, “Well, we have changed our mind. We will develop, test, and maybe even deploy round one, but we will never deploy en masse on that platform.”
LM: Do you think Linux will ever be the dominant OS on the mainframe?
SK: No, it will never dominate. There’s too much legacy; there’s too much in terms of applications that have been built; there are too many sophisticated mission-critical systems still. MVS is MVS is MVS. It does things that other things just can’t do. It’s just a fact of life. While there are many examples — Amazon’s one — of high-end transaction providers, you know, consumer companies who are running Linux, if you want to know transactionals, you’ve got to go to a SABRE or a Charles Schwab and sit there when the New York Stock Exchange opens to see real transaction workloads. It’s a whole different league.
If not for the mainframe, I don’t think some of the companies doing work with us today would have even gone to Linux. And that’s its back door into the enterprise. What it’s going to do is get them to Linux, and I don’t think they will stay on the mainframe forever. I think they will become comfortable with the platform, and [then] they’re going to go off somewhere else.
LM: What do you think of the GPL? Do you think it’s contributing to the autonomy of the Linux platform?
SK: I think it does, but the platform is still — in commercial terms — too young, too immature to figure out where it’s going to go. We’re seeing the beginnings of a whole new movement here. I don’t think it’ll be as big as the Internet, but it could be meaningful in a different way.
People say you can never have a proprietary extension to Linux. That’s BS. Of course you can. Nothing prevents IBM from coming out with add-ons to their Z platform in microcode that makes Linux do different things. Sure it’s completely permissible; there’s nothing wrong with it. So I think it’s too early to tell where it will end up.
The more competition there is, to some degree, the more pressure there will be on the vendors to do unique things. To some degree customers want competition in the marketplace, but the more customer competition there is, the more vendors have to differentiate.
Open Source Plans?
LM: Have you released open source code?
SK: Yes, in two instances. They are both extensions to our management products.
LM: What license did you use?
SK: The GPL.
LM: That seems like an area where there’s a difference between CA and IBM. IBM’s released a lot of Open Source code and they keep moving into these new spaces, with Eclipse for example. In your LinuxWorld keynote, you intimated that more community things would be happening at CA. Can you tell us a bit about them?
SK: We will never open source everything that we’re doing on this platform. I’m really clear about that. That’s just the way it is. I will also never go to the other extreme and say, “Nothing will be.” I think there are a whole bunch of things that make sense. The problem is, you’ve got to be careful in what you do. I make security software. It runs in a hairy space, right? I would be doing an absolute disservice to turn that stuff loose. My customers would just go nuts. They would absolutely come unglued if I did that. I do believe that there are pieces of that technology that do make sense to be out there — where we can get other people to help us — but we have to re-engineer some of the base technology so I can separate them.
LM: So in what areas do you think it makes sense for CA to open source software?
SK: I think on the management software side, the agents that manage the platform. The network element technology, for example, that’s rapidly changing. That makes sense. There are components of the database products that make sense. I would never turn loose the optimization engine, for example. That I consider truly proprietary. But our Ingres database product was developed in a university community environment. I don’t have a major problem, in principle, for it to go back there. But the optimization engine, which we have invested 20 years in building, I have a problem with that [being open sourced]. That’s my competitive differentiator. There ought to be a clean way to talk to it, there ought to be published standards, I ought to make those standards open, and I ought to say, “This piece is mine, but all these other things I’m going to turn loose.” But I have to get the technology [set up] in a way where I can unbundle and separate them. If not, I create a big mess.
CIOs AND Open Source
LM: We’ve been talking about Linux a lot, but to what extent do you think the phenomenon is really about open source, the development methodology?
SK: This is probably the statement that will most disappoint your readers. I don’t think it’s got much to do with that whole open source thing right now. The space that we’re in, it’s not about open source. It’s about customers coming to us and saying, “This is a terrific platform. We never realized it could do the things it could do, either from an embedded side, which is completely customizable, or from the really high-end side. There’s no other platform, short of going in really haggling, begging and pleading with the vendor, where we can get it to do that.” If people came to us and said, “Can you customize Unicenter and make it this big or this big or this big?” I’d go out of business if I did that for every customer. So a big driver today is customers saying, “I didn’t realize you could do this.” There’s this surprise factor. “I didn’t realize that it’s stable and IBM has embraced it.” So to some degree, the purist view is not what’s driving it. Ultimately, it’s an attempt to save money.
LM: So does the open source nature of Linux interest your customers at all?
SK: At the CIO level, no. For them, it’s stability, price performance, efficiency, viability, who else is doing it.
LM: And they couldn’t care less that it’s open source?
SK: I know it’s going to disappoint a lot of people, but no. Two levels down in the organization, it makes a huge difference.
LM: How many Linux distribution vendors do you think there will be in five years?
SK: Less than today. I think consolidation is good because it will do a couple of things: It will make the ones that survive the first round — and we’re still in the first round — stronger. It will send a powerful message to customers that consolidation is permissible in this marketplace and viable. Right away people think, “Customers like bigger things.” Not necessarily. Customers also want to know that the guy who can survive is valuable enough for somebody else to absorb. That’s critical in this day and age. It was not critical two years ago when the market was all up and flying, but today it’s critical. A customer wants to know that XYZ Company that I bought stuff from is viable [and] even if they can’t make it, that somebody’s going to pick it up, which makes the platform look and sound safer. I think if the big hardware guys become providers, distributors, supporters, that’s important. Customers look to them naturally. That’s where the platform has come from for as long as we’ve known computing, with the exception of Microsoft.
LM: So are you saying that you think IBM should do its own distribution?
SK: Well ultimately I think that customers would love for them to.
LM: They don’t want to do that, though.
SK: I know. Customers would just love for them to.
LM: And would that bother you?
SK: It wouldn’t bother me one bit. I don’t want to be in the distribution business, because I’m platform-agnostic. With Linux, you have this — it’s not an issue, it’s not a quandary — but you have a particular circumstance where you have a community that feels that they own the platform and then a bunch of capitalists have shown up around it. Without the capitalists, the platform ain’t going to go anywhere. And without the people under you who have done it, the platform never would have gotten [to where it is]. So you need both to get along, and the question is where is the happy medium. The two halves of the community have to become one. Without one, the other can’t survive. People are very hung up about what one’s interest is over the other. It really doesn’t matter, because as far as I’m concerned, if you don’t have both, it just won’t work.
Robert McMillan is Editor at Large for Linux Magazine. He can be reached at firstname.lastname@example.org