Looking Ahead Greg Papadopoulos, CTO, Sun Microsystems

Sun Microsystems CTO Greg Papadopoulos has held several positions at Sun since joining in 1994, but his pre-Sun resume is also impressive. He co-founded three companies, including PictureTel (videoconferencing), Ergo (high-end PCs) and Exa (computational fluid dynamics); held engineering positions at Hewlett-Packard and Honeywell; and was a professor at the Massachusetts Institute of Technology. Papadopoulos recently sat down with CRN Editor in Chief Kelley Damore to discuss future technology and its implications for customers and the channel.

PAPADOPOULOS: As we go through this next decade, and five years being in the middle of that, we are going to continue to build out and expand the kinds of things that connect the network. So we've moved from getting all these computers connected,servers and desktops,to now this wave of cell phone technology. And the next wave, and we'll certainly be in the thick of it in the next five years, is the Internet of things.

[These things], ranging from RF [radio frequency] ID tags tracking inventory in realtime to lightbulbs to whatever will get made or sold, will have some sort of network.

And all of this puts two forces of very large pressure. There is a pressure on the developer to be able to figure out how you write software for this world. And then there is the pressure on the operators or the system administrators on how you project services within a cost constraint. So those two communities are under huge pressure now to follow and to be able to bill out essentially what those functions will be off the network.

So Web services appeals to the developers. Basically, it takes the developer, who for the past 20 years has been writing essentially shrink-wrapped software. The revolution that we've lived though in the last decade [involved] developers writing for the PC or the Mac. Now we have developers facing the challenge of, 'Oh that's not the point anymore. I don't see the bits. What I do is create a service on the network and have people go to that service.'

So the developers have to wrap their heads around what it means to develop, and then the people who are running the services need to [worry about] getting the service delivered.

So the short form is that Web services is our next step in the evolution of how we write software for the Internet. I emphasize it is the next step because it is not the end. It is a path along the way, and it is an evolution and the basic ingredient of Web services. So in the five-year time frame it will come under a lot of pressure in its own right: Issues around the Web services model, is it really the right model for the next wave of the Internet?

CRN: Why so and what type of pressures?

PAPADOPOULOS: Web services are focused on the notion of delivering Web content and transactions from the centrally managed client down to users. Content is a big data center. That model is a fine one for the applications we are doing today. But as the network scales, you'll want increasingly to talk to one another directly,peer to peer. So you'll be writing services to be assembled in a very distributed way across the network. Even if you replicate them and do load balancing, you won't be able to scale up enough.

So if I say, for example, what is a storage service for the Internet? When we think about everyone wanting to throw their documents and digital photos onto some service on the network, it isn't going to be done on some big monster Web service that is hosted by someone. It is going to be many, many people participating in a storage collective in a distributed way. And that is why at Sun we are really out there trying to drive and clarify an open model around Web services. We're also investing [because we think the industry] will be far more distributed.

CRN: There is also a need for interoperability. Do you think the practice of introducing dueling standards with one winning out will continue? And will there be a Web services standard in the next three to five years?

PAPADOPOULOS: Everyone should insist that the standards emerge and coalesce. This gets into what I think is a very important point and gets into people in the reselling and channel business.

The one thing that is great about the Internet is that there is a set of open standards and interfaces that everybody builds on and everyone owns. . . . We have a very strong push for royalty-free in standards. Companies should not benefit from [becoming] a standard.

There is a lot about what we as a community should insist upon because as long as systems become more and more distributed, so should the innovation. And what has made the Internet light up and go as far as it has is that is has been very distributed. Anybody can put up a server and Web site and become the next eBay if they have the right app. And we want to preserve that going forward. I think that is really at the core of what makes Sun and how we think of the technology moving forward. We lived it for the last 20 and want to live even more deliberately for the next 20. This has to be full community where anybody comes in, with their particular skill set, and is able to build and extend upon what the core structure and values are.

CRN: How do you delineate what you provide as Sun and what a solution provider can do to build on top of that?

PAPADOPOULOS: We are in the general-purpose infrastructure business. We want to be building things in an open and valuable way that the experts and specialists [can build upon.]

We're not a service provider. It is very much focused on our core values of providing the most reliable infrastructure. Now what happens as we move forward, which is almost always the case, functions that used to be specialized or value-add become sedimented into the infrastructure. Years ago, we had a really big business selling Internet stacks into Windows systems. Now everyone believes that should become part of the operating system just like Web servers became part of the operating system and in the future so does application servers. So there is this process of taking waves and patterns and common sign elements and bringing those down into the platform and the infrastructure. So the very important thing for anyone participating in the value-add process should always be looking at where are the systems going and where the next layer of value is going to be.

CRN: How do these technologies change the way in which businesses operate?

PAPADOPOULOS: I think networking, at the core, takes time out. Information becomes much cheaper, and it actually lowers the barrier to competition. The other side of it is I can be a business that is more efficient in networking and information flow. It also means that my competitors do it, too, and consumers have more [flexibility] to move from supplier to supplier.

People really have to dig down into what their core value is. We have to do the same thing as a company.

CRN: Is it more pressing as you offer these services over the Web to ensure security and privacy?

PAPADOPOULOS: It is a core aspect of why a customer will end up valuing a partner,whether you are a service provider or [another type of partner],a lot of it will have to do with the trust. We have the technology to preserve privacy, to make operations secure. What we need to do is be systematic and have the processes around it and drive agreements.

Liberty is about being able to peer different identity systems that people have. So if you look at a customer database of an airline and a database of a calendar service [enabling an] airline reservation to automatically pop up on your calendar. But that doesn't mean either company gives out their customer list. Both companies need to preserve the level of privacy that they promise their users.

CRN: If you look back at the dot-com explosion, what lessons were learned and what were some of the biggest mistakes?

PAPADOPOULOS: At the core was this recognition that happened across society that networking and the Internet did something fundamental to the economy and the way we live. That was the recognition, and it is still true. It is more true than it was a few years ago. The network has grown and made progress. The motivation was fine, but the way that we expressed it was that all of our rules of economics get thrown out the window, and we begin valuing things that don't have concrete business models behind them. Those were the mistakes that were made.

CRN: You mentioned peer to peer earlier. Why is distributed computing taking hold now, since the notion of it has been talked about for years?

PAPADOPOULOS: Distributed computing has been talked about as long as I have been a computer scientist, and that is going on 25 years now. The pressures that come from scale.

Truly, you get to this point where it is hard to think about how I will support devices in a centralized way. There has to be something about how the device joins the network and brings a little bit of resources and how I am doing the network in a fractal way. We will end up with larger-scale networks because we can't make the other stuff scale.

CRN: If you were give advice to leading-edge solution providers in terms of technologies they need to be investing in now or particular industries, what would those be?

PAPADOPOULOS: At the systems level it would be how we do virtualization. All of what is happening in storage virtualization, in network virtualization.

I would say [VARs] need to have a keen eye on what is happening to any computer that touches the network. In the past 20 years, we have designed machines that are very processor- and memory-centric. We haven't really designed computers where the network is the center of the machines. To put it another way, how our computers will be built out of networks. That will change fundamentally computer design and processor design, and you will see that happen in the next three to five years.

In addition to virtualization, I would be very focused on what is happening at the edge of the network. As I step out into markets, I do believe there will be an interesting resurgence in the telecoms related to the network. I would also be looking and trying to follow efforts converging the identity model and billing.

The emergence of RF tags will cause a lot of shaking up in retail distribution. Go look at what RF tags will do to Gillette. Very interesting things will happen.

CRN: Can you talk about the possibilities of nanotechnology?

PAPADOPOULOS: In the long term, absolutely. In silicon, we're getting down to the atomic scale and pretty close to that with the structures we build on chips. The electronic layers are only a few atoms thick.

Nanotech is a big leap to try to do molecular assembly. That will be where engineering will eventually end up, because we are along that path. Whether it comes out as nanomachines or nanobots, I am rather dubious.

I think [we will see nanotechnology] in materials,so how we use our ability to manipulate at a molecular level. The other area will be in molecular manipulation of DNA and the genome, and that is something we are doing today. A lot of the nanotech folks gloss over it. I think that assembly of the micron biological level is going to be the one that makes the most progress.