 |
| Rob Gingell |
He has spent much of his 17-year career at Sun Microsystems trying to get the other technology gurus at the company to follow his lead. As the chief technologist for Sun's system software group, Gingell ran herd on Solaris, Java, and the entire portfolio of servers and development tools. Four months ago he was appointed Sun's chief engineer, and now is responsible for crafting a cohesive strategy as Sun moves from it first-generation systems based on Unix to a second generation oriented around Java.
Gingell talks about his desire to open source Solaris and intermarry it with Linux. He also discusses his focus on other parts of the software stack, especially Java, and why he believes Sun will succeed at a time when Solaris and SPARC are no longer the company's crown jewels. Get an inside look at Sun's strategy in this first installment of the two-part interview with Gingell. When you're done with Part I, be sure to check out Part II, in which Gingell talks about how he thinks history will repeat itself--to Sun's benefit.
Tech Update: As Sun's chief engineer, what do you do that's different from the other technology chiefs in the company?
Rob Gingell: My charter is conceptual integrity. Until a couple of months ago, we never had a chief engineer, so this job is different in that respect. Prior to that, I was the chief technologist for the software systems group, which included Solaris, Java, the iPlanet products, and the development tools.
Within Sun, we have a bunch of chiefs primarily because the structure of the company is a little recursive. Every manager of a large staff has their person responsible for representing the technology interests and portfolio of that division. As a software chief technologist, my job was portfolio management, but I did the architectural stuff as well, so I was fairly unique in that regard. Yeah, we do have a lot of chiefs. I've been on a lot of annoying panels where you start to wonder if any one is doing any work with all these chiefs.
Tech Update: What do you mean by "conceptual integrity?"
Gingell: That's a short description that I use. If I'm successful, then when customers buy a stream of our products and slap them together, they ought to be working. If it happens that they slap them together and they don't work, or they stop working, or work in unexpected ways, that's probably a failure of architecture. At some level, I need to figure out why that happened and make sure we put things in place to put it back together.
My goal in life is to make sure that all the brains in all these buildings [at the various Sun campuses] are effectively employed and create as much as they can. If only one person creates the ideas, you only get one person's worth of ideas. I'd much rather have 30,000 people's worth of ideas. It's always much more powerful, although you have to deal with the arbitration between the conflicting ideas.
Tech Update: Company officials that I've met with in the past have talked about how running Sun was like herding cats, with a lot of diverse interests running in different directions. How much of what you do is focused on keeping the company going in one direction so others can see what the mission is and see what the future is like?
Gingell: A lot of it is like that. I actually hope that it's never true that the herding cats phenomenon vanishes from Sun. Some of the chaos you're referring to is what makes us interesting and vital, and keeps us from getting locked into a "we're doing this because we did it last week" mentality. That level of chaos, while it's annoying at times, is also fairly powerful because it's the product of having all those brains usefully applied. Where it's a negative is when you have no way of arbitrating the chaos. That goes back to my arbitration role, which I did locally in the software group for many years. It's a new scope expansion to consider doing it for everything all at once.
If I'm successful, we'll more efficiently surf the froth off that chaos, mine it more effectively, and more quickly translate it into "OK, this is where we are going and how that idea over there contributed. Next idea, please."
Tech Update: Where you have this chaos and you see it as a positive, your customers certainly don't necessarily feel the same way. Is there a disconnect?
Gingell: I haven't personally run into that many customers who are confused about what we're doing. Some of the publications are more confused about what we're doing than some of the customers are, although I certainly don't talk to all of our customers. I've been at Sun for 17 years and I haven't woken up on any day confused about what we're doing or why we're doing it. What is going on is there are a lot of people at Sun who have not been there as long.
There's a lot of primate behaviors in any large organization. The things that everyone works on--those trees that they're staring at a lot--are sometimes confused for the forest. I won't predict that if you talk to a random selection of employees in the hall that they'll all tell you the same thing, but I'll bet most of what they'll tell you can be mapped to the same essential thing.
The way I think of it is that we're moving from our first generation of systems to our second generation of systems. Our first generation of systems was designed to run the Unix application base, and the second generation of systems is designed to run the Java application base. They incorporate the Unix base into it, but that's a definite shift in the structure of what our products are.Tech Update: Is the operating system becoming a commodity, and no longer a differentiator between Sun and its competitors?
Gingell: Yes it is. Twenty years ago we added value to the world by integrating a bunch of components that were largely defined by the OSI [Open Systems Interconnect standard] and operating systems and things like that. Now, the world expects that as a given. It's not a differentiator or special thing.
So what is it that a systems company integrates in the year 2002 to create value? Mostly, it's integrating a bunch of components with IP addresses that deliver functionality.
The whole business around software stacks is really about functionality. All that business about creating Java and the Internet and so forth was a setup for building discreet [uniquely addressable] software entities that live on the net, and now is the time to exploit the fact that Java has been relatively successful. The Java developer base is an order of magnitude larger than our Unix base. We made a $12 billion business out of the Unix base. What can we make out of a developer pool that's ten times the size of the one we have for Unix? Well, I don't know if it's a ten-times larger business, but it's a bigger one.
Tech Update: So, in a world where operating systems and processors are commodities, what differentiates Sun in a way that IT managers should be drawn to you versus other solution providers?
Gingell: Sun is the only company besides IBM that has the engineer resources and does processors, OSes, tools, and applications. That is true. We grow by building things that pursue the new Java apps base that we've spent most of the last decade helping to promote. We [Sun] are the last people standing except for IBM who can bring all those skills--an apps base, chips, ASICs, system software, languages, applications, and system buses--to a problem and create answers that employ all those skills. IBM doesn't seem to want to bring any one skill except for the services skill to the party, which is inherently a very costly approach because the humans are always more expensive than the parts.
If it's a war of humans, well, we're one-tenth IBM's size. It's like, "When did we want to get beat up?" would be the nature of the battle. On the other hand, if everybody has knives and daggers and somebody shows up with a Gatling gun, you don't need that many people to run the Gatling gun. That's an unfortunate analogy, but if you need that many people, technology always outdoes people in the end. So, as a products-based company, we are focused on what technological offering can we create that fills this picture.
Tech Update: You are now making a push in the Linux direction. As a part of Sun's brand equity, would it be a mistake to say there's a quality difference between Solaris and Linux?
Gingell: It would be a mistake, but I don't think that's what we're doing. Maybe I'm more radical about this, but I'm on record as saying I wanted Solaris open sourced. The reason I wanted it open sourced is that the Unix community has existed for many years. It's only been lately that we label it as the Linux community. It existed in the Bell telephone days in the 1970s, and the BSD days in the 1980s, and it got diffused in the 1990s as a result of a bunch of the balkanization--the Unix wars. But there's been an underlying community that contains all of the people that I know They're still hacking on OSes and for the most part, they're hacking on Linux because that's where the center of brainpower has moved.
Tech Update: Are you saying it's only a matter of time before that community gets Linux to the point where it's up to any task that Solaris is capable of? What is the rationale for open sourcing Solaris, rather than focusing on Linux?
Gingell: The consequence of the Unix wars was that all the elements of that community got fractionalized, partitioned, and balkanized. At some level, Solaris was sort of the winner. As a result we're the last remaining balkanized Unix. Our developers really don't relate to anyone else.
On the other hand, we've already done things that everyone else needs and wants, and it's sort of a waste of time to have everyone else spend the same millions of dollars and thousands of lives on doing scalability and that sort of stuff. So why don't we open source the thing and see if we can get the tribes to intermarry and re-accomplish this community thing? Maybe we call our thing Linux by Solaris or something like that.
The value of what we have is not really the code as much as it is the body of people we have who do the work we need done on a predictable schedule. That's always been the real value of the Solaris group. What we want are the qualities that we by and large invested in the Solaris code base to become a part of the bigger intermarriage. Five years from now when all the tribes intermarry, who is going to know what's Solaris and what's Linux, and who's going to care? Nobody's going to care because, for the most part, we think they'll be running Java-based apps anyway. So the OS is a thing that gives you reliability and administrative power and things like that. But our real value is in having a group that can operate at the OS level in the applications stack.Tech Update: Given that the outcome that you're seeking means that you want some of the same quality attributes found in Solaris today to be a part of Linux, the software layer, especially the OS, ends up insulating the hardware layer. What does that mean for SPARC? At that point, you have this combined thing that commoditizes the hardware.
Gingell: Right. In some ways, that's really what Linux is. It's the cost reduction commoditizing of Unix over cheap volume-produced hardware enabled by the PC marketplace. Part of the answer to that is that if everybody is invested in Java Virtual Machines, then they are not invested in SPARC. But neither are they invested in Intel or in any chip-based instruction set. That's a device that is to the system of 2002 the way resistors were to the systems in 1982.
If you stand back from the Unix industry, you'll notice that the apps space has leveled off for many years. People may be doing revisions of things that already exist, but you don't hear about a lot of new applications. Linux is having a lot of growth primarily because it's just the existing apps moving over. Somebody issues a press release that says, "I did that port." There are thirty years of Unix history for Linux to catch up on and once it's caught up on it, then what? All that activity is taking place somewhere else. It's occurring in this networked environment rather than in the boxed-space environment.
Some of those devices may differ, however. For example, some chips allow you to build low-power devices that run at a certain price point or go at certain speeds. Or, they may consume a lot of power, but they give you the ability to have a really haul-ass, vertically scaled database processing thing. What's valuable to us is that we have one-eighth of the world's CPU designers, not the instructions that they work on.
It's the fact that when I build a data residence device that some people might call storage, I can have processors, ASICs, and my own systems software in it. [At Sun] I have all those skills to build the perfect data residence device or something that fits the market definition of such a device. Let's just deliver the software components with the things that are needed to power them so I don't have engineering arguments with people saying, "Well, this management agent has a footprint that's too big." Why don't we just deliver the hardware with the management agent so we don't have to worry about it? Well, we can make the chips and the software and all that stuff, so, we can actually do that! Right? Nobody else can do that. That turns out to be a pretty useful value proposition to a lot of customers.
We're not telling people to delete it and we're not saying to rewrite all their Unix apps in Java or anything. In fact we expect that world to just sort of surf along at whatever level it's at.
Read the Part II of this exclusive Tech Update Unplugged interview.
