Frye: Curing corporate Linux phobia
Published: 05 Dec 2002 08:29 GMT
Four years ago, Dan Frye convinced IBM's management it would be smart to jump on the Linux bandwagon -- a wise suggestion. Not only did that decision afford Big Blue a hipper image, but it also helped the company open doors that otherwise might have remained shut. A soft-spoken man with a doctorate in atomic physics, he now finds himself a spokesman for open source inside the world's biggest computer company.
With IBM putting billions of dollars behind the Linux operating system, Frye helps Big Blue chart its strategy for coaxing the young operating system into business. But as head of the Linux Technology Center, he must strike a delicate balancing act. While helping direct IBM's Linux strategy, Frye also must avoid the appearance of trying to impose too much order on a famously independent community of outside programmers. He is aware of the risk, but nonetheless believes the corporate world is losing any lingering hesitation it once held about adopting Linux. If he's right, then the good folks at Microsoft have something to worry about for 2003.
When you're talking to somebody who's leery of Linux, what red flags do they raise?
We've been talking about the myths about Linux over the last couple of years. The myths are that it can't be secure, Linux doesn't scale (to large servers), Linux doesn't have applications, and there are no Linux skills.
Some of those have gone away. No Linux skills? Customers don't ask about that anymore. Linux can't be secure? That's still something of a hot button, but less and less.
Linux doesn't scale is something that is still a concern, but as Linux begins to show better scalability on larger SMPs (symmetrical multiprocessor servers -- higher-end machines), that's beginning to go away as well. (Editor's note: Linux now works well on four-processor machines, and experts believe eight-processor servers soon will be able to use the operating system. It will take longer to catch up to Unix, which is common on systems with 32, 64 and more processors.)
I think the key legitimate concern is that customers are no longer asking if there are any business applications available on Linux; instead they are rightfully asking: Is my business application available on Linux?
Which red flags will fall down in the future?
That one (of not having enough applications) will fall down. The number of applications on Linux continues to grow rapidly. We don't know of any fundamental red flags that won't come down.
So what barrier will prevent people from using Linux on every single computer they buy 36 months from now?
Well, Linux, like any operating system, can't possibly be right for everybody. And for some time, Linux is not going to be capable of doing those very high-end, large SMP (symmetric multiprocessing) functions.
Three years from now, will Linux work well on a 32-processor server?
It'll be right in that space. The next release, 2.6, will not scale well on 32-way servers. The release after that, available in 2006, could begin to approach 32-way.
Do you think the Linux development process will become more and more formal?
It already is becoming more formalized. It will never be the same.
Is this something the corporate world is driving?
I would imagine IBM is very interested in being able to achieve certain milestones by certain times. It's more accurate to say it's an evolution we are supporting as opposed to driving. We actually have learned in the Linux space not to be dependent on those specific dates.
It seems to me in the last four years Linux has become a lot more controlled by the 200-odd people in the Linux Technology Center and by colleagues elsewhere in the established computing industry.
We actually think of that as a positive, that a large majority of full-time people in Linux are paid. They're paid by a variety of different people, whether it's Red Hat, SuSE, Conectiva, IBM or HP. We consider that a positive because it promotes stability and gives a measure of stability that a customer can understand. The technical self-interest that the community has is now aligned with everybody's day-to-day financial self-interest.
How many members of the Linux Technology Center are working on IBM-directed projects, and how many of them are Linux guys working on what captures their fancy?
It's close to 100 percent of the former category. Some have personal-time things they work on. Even though our mission is to make Linux better, it's not a random mission. We ask certain people to work to improve SMP scalability, or improve network security, or improve the standards. We don't hire them to do what they want. In some cases what they're doing is why we hired them. We went out and hired the PowerPC maintainers and kept them working on PowerPC.
At a corporate view, we have the same right as everyone else. It's the scratch-your-own-itch philosophy of the open-source community. We're scratching our own itch. We're working on the things important to our own customers. The community's been very supportive of that. They understand that.
Is the Linux community becoming more buttoned-down and corporate?
I don't think so. I don't see the inclusion of better planning and better tooling as the same thing as buttoned-down and corporate. We still think Linux development is going to be freewheeling, innovative, fast-moving. We don't want to stop that.
When will Linux servers outship Windows servers?
IDC has some data that shows it as some time in the four-year time frame.
IBM has an army of attorneys and probably the computing industry's largest business in licensing its intellectual property. Are there going to be some legal challenges to the General Public License (GPL)?
I don't know. We take our obligations from all licenses very seriously. I think there are some misconceptions about Linux and the licenses. We are pretty clear in telling customers that their applications cannot be infected by the license of something else. Certainly IBM has a very large non-open-source, proprietary software business. That software runs fine on top of Linux.
What areas are ripe for open-source software?
Open source is one excellent way of providing IT infrastructure -- networking, Web serving services, base application development tooling, client desktops. The things that are generically used by everybody are good candidates.
The community is large enough with enough varied skills that you can overlap between the people who want to make the software better and who have the skills to make it better. As you get into the middleware and higher up the stack, that becomes harder to do. The people who want databases to run better don't necessarily have the skills to make them better. That's especially true in the application space. People who want to make hospital admission software better -- the people who use it -- by and large don't have any software skills, and they're not likely to form an effective open-source software community.
How well will Linux eat into Microsoft's market share for desktop operating systems?
We think it's unlikely Linux will eat into it, with the current view of what (PC) clients are. But that is going to continue to evolve as well. If you think of the business client of the future, it's going to look much more like a portal with back-end functionality driven off a server somewhere. In that space, open source may well be a very viable alternative. But we don't see Linux running on every 8-year-old's computer.
Do you see the end of the split between the KDE and the Gnome desktop software projects in the next three years?
I would certainly hope so. One of two things will happen. They will figure out a better way to interoperate, or one of the two will pull away and become the winner. The ecosystem is large enough to support two development teams but not large enough to support two uninteroperable solutions.
Do think that will be bridged in the next three years?
I think the market will require it. Either there will be a winner or they'll become interoperable.
Will IBM produce its own version of Linux in the next three years?
We have no plans to do so. We're very happy with our Linux distribution partner strategy. But we'll do what it takes to be successful in the market.
What's the single biggest project you hope to attach to the open-source community in the next few years?
SMP scalability is a major one, networking functionality is second. Serviceability tools is a major one. And standards -- we're probably the biggest pusher of Linux Standard Base.
When you say networking, do you mean IPv6 (the next overhaul to the Internet Protocol)?
IPv6 and SCTP (Stream Control Transmission Protocol) are the two we've spent the most time on.
Are those the ones you believe you'll be able to check the check box in the next three years?
Oh, absolutely. We will check the check box on SCTP when 2.6 comes out. It's looking like IPv6 will make it into 2.6 as well.
In next three years, where will Microsoft still be ahead of Linux?
I spend just about zero percent of my time worrying about Linux versus Microsoft. I spend all my time worrying about how to make Linux better and what customers need. I don't keep a checklist of what Microsoft has and Linux does not. That might sound like the canned corporate answer, but it happens to be true.
It does sound like a canned corporate answer. Here's my perspective: I haven't run into an operating system person yet who doesn't compare his or her operating system to the other operating systems out there. I would be stunned to hear you don't pay attention to the other operating systems.
We absolutely compare ourselves to Solaris. That's the target. We want Linux to be, for example, just as good or better than Solaris for the telecom market.
Have your say instantly, and see what others have said. Go to the Linux forum.
Let the editors know what you think in the Mailroom.
Did you find this article useful?
11 out of 23 people found this useful
Digg
Slashdot
Del.ici.ous
Stumble
Reddit
