Turn your computing power on and off like you do your faucet when you're thirsty. And pay only for what you use.

Related Items »   NC05Q2 recap »   Grid solutions from Sun »   eBay and Sun »   More data center simplification

No longer are economies of scale and innovation mutually exclusive. Despite utterings of industry naysayers, innovation is far from dead. Jonathan Schwartz, Sun's president and COO, and Andy Ingram, VP of Marketing Scalable Systems, spoke with Sun Chief Researcher John Gage during Sun's latest quarterly NC event. The future of enterprise computing and specifically the data center were the topics of discussion. Below is a snapshot of those interviews.

John Gage (JG): The computer industry is changing fundamentally in how it responds to the needs of its customers. How do you see this evolution?

Jonathan Schwartz (JS): There are two ends of the spectrum when we look at how the industry is being viewed. One viewpoint comes from certain analysts who say that the industry is mature, all the innovation is gone, and it's really kind of over. Sun's view, and that of many of our customers who use technology as a competitive advantage, believe we're just at the beginning of a sea of change in terms of how technology is being adopted and used.

If you look at the history of any technology and its evolution, you see examples. My favorite is the electricity marketplace, which when it was introduced was very esoteric, very expensive, and could only be deployed by one individual — J.P. Morgan. He was one of the wealthiest people in the world. He had a generator in his backyard and electricians who managed the generator, and when they went home at night his lights went out. Over time, we evolved to a standard called AC power, and there were a couple of companies — General Electric and Westinghouse — that made the fundamental engines that generated this electricity.

Now we don't think much of it, but we use electricity every day. We don't even think about how much it costs because it's so cheap. Well, every technology goes through those evolutions of customization and high expense to standardization and consolidation, and then ultimately the delivery over a grid. I think what's happening in the computing industry is that we've gone through the era of customizing everything. We've moved toward an era of standards, and whether they're industry, software, or platform standards, people are for the most part consolidating their investments now. The really interesting innovation is how to use that consolidation to deliver a service.

JG: In these consolidations lie economies. With a massive change in efficiencies and economies, won't we see a huge change in the price points?

JS: We've seen the price points in the technology marketplace dramatically plummet. And that has opened up a vastly larger marketplace, as happened in the electricity marketplace. The example I love to give here is the fact that six years ago, what was the market for a search engine with one text field? About zero, right? Because no one would buy that product. Yet today, literally hundreds of millions if not billions of people are searching online using the simplest search tool ever delivered. Why? Because they could bring together the combination of very, very low price — free — along with trivial accessibility, and ultimately, the economies of scale that allow them to dis-incent people from recreating the wheel. And the rest of us have access to the easiest search service available in the world.

The same thing is happening inside enterprises now, as they look to determine whether they should build something on their own, buy a product off the shelf, or perhaps just use an online service if it offers the best economies. As that occurs, it changes the economic landscape, but it also opens the market and creates more opportunity for everybody.

JG: How do you distinguish the different grid services that could be provided to customers?

JS: A variety of grid services are already available. eBay is a wonderful example of a commercial grid — an online auction. It is ubiquitously available and has a very attractive service model. There are already large-scale grids in existence off PCs — the telecommunications grid, American Express, Visa, MasterCard. Those to me are all identity grids. They're basically asserting that they can vouch for you and your ability to pay for a product at some point in the future.

What we're beginning to do with our grid though is this: It's not just about network services, it's also about network capabilities. When we unveiled the Sun Grid and said that for a dollar per CPU per hour we'd make the largest super-computing grid available in and to the world, that began to change the dialogue among some of our customers. They now ask themselves why they would build their own grid, which from a technology standpoint may or may not provide a competitive advantage, when they can go use someone else's that will? People have started looking at what their data centers were costing them in terms of CPUs per hour. And those who find that they're paying more than a dollar per CPU hour now have an option to reduce their costs by utilizing the network service provided by Sun called the Sun Grid, which is available on a worldwide basis.

JG: So in a way, this is a new engine of growth in the high-technology industry.

JS: Fundamentally, Sun has been investing in network computing since its inception. I don't think we're in any way distracted by those who believe that the industry has matured. I think if anything, we believe that now is the time to really double down on the innovations that will cause the transformations to occur. These transformations won't destroy value. They will create value for those that are in a position to take advantage of innovation. Our most exciting time is ahead of us.

JG: And on to the data center. Andy, how do you see the evolution of the data center over the next five years?

Andy Ingram (AI): The big change is going to be in moving from a highly-customized environment to more standardization; and ultimately, the ability to utilize a utility or master aggregator. But really, in the end, it's about driving economies of scale in the data center.

JG: How do you achieve higher capacities and economies of scale without adding complexity?

AI: That's the very definition of scaling — the ability to add capacity without complexity. A way to think about it is if you write your application to provide a business benefit, to differentiate in the marketplace, lower your cost, or gain competitive advantage, then more applications are good. However, in the data center, more different types of platforms are bad because they add complexity and complexity costs money. You want to aggregate as many applications as possible onto as few platforms as possible. That gives you economies of scale.

JG: How can a Sun customer achieve this goal? Are there tools, experience, or some form of knowledge that Sun has that can help them reduce complexity?

AI: There are two things. First, what is the basic path? Second, how can Sun help? The basic path begins with the application, so that rather than writing applications uniquely, with customized and differentiated platforms for each, you write your applications in a way that allows you to aggregate them. Then, there are techniques you can use in the data center that allow you to aggregate your different platforms together and drive for identity, security, continuity, and everything that goes along with that.

Take portals for example. Perhaps you have a portal you've written for your customers, and a different portal for your suppliers. Both have very different applications in terms of the communities they're reaching out to and the kind of service they provide. But underlying that, the middleware, the basic structures are in fact common. So, rather than writing them in a way that forces you to have two different platforms, you put them both on the same operating platform so that the middleware, the tools, the very techniques of developing them, allow you to have similar behaviors between your portals so that you can put them onto one infrastructure and share the resources.

Before you might have designed a data center for a certain pattern of peak load. By combining several different applications with different peaks, you have far greater efficiencies in terms of providing those services using your invested capital. Plus, if you want to roll out a new portal, you just add capacity. You don't have to rearchitect, so your time to market is much quicker.

JG: Is there an area people should focus on? Can you give guidance to people trying to decide what they should look at first in their environment?

AI: We've been working with customers who are on the leading edge of driving for Internet architectures. We've seen best practices, and we basically have built up a set of offerings for our customers in terms of services, reference architectures, best practices, products, tools, and standards to help them make this transition. We can help customers build the roadmap to this better place. It starts with the application, so we have expertise, we have tool sets. We're driving standards like JSR-208 to help make that economy of scale possible by changing the behavior of the applications. Then we can help you with the infrastructure to consolidate what you have today, to architect your resource pool, so that you have the minimum number of operating platforms to drive for continuity, security, identity, content management, and governance. We'll help you set up a structure, manage it, monitor it, and help you determine a strategy for what you do in-house versus what you outsource to a master aggregator. All of these things are a part of the process and journey to where data centers are going.

JG: Five years from now how do you see the destination?

AI: It's a matter of having options. Customers will have aggregated applications into their infrastructure, will have gotten these economies of scale, and now they'll have a choice. They can choose to outsource the things that aren't critical to them. They can have the things in-house running much more efficiently, they can deploy new services more effectively, and they can focus on taking data, turning it into information, and driving their business.