ENGINEERING
Dell exec: Standards, simplicity are key
How do you envision the computing center and the networked infrastructure supporting it to change in the next couple of years? As standards continue to emerge you will see more and more Intel and Microsoft/Linux-based systems in the data center, and a continued decline in the Unix space. That's obviously advantageous to us. Many of our competitors have big bets in the Unix space, so we believe that that's going to continue to deteriorate their performance and be a drag. At the same time, customers are moving more to standards-based products. We only have those, and all our efforts of development and customer satisfaction are based on those products. We see that the move in the industry technologically is coming toward us. 
How does Dell use standards to commoditize a market and capitalize on it? If you look at where standards play, they don't play very well with companies that have proprietary structures. They don't have a cost structure in terms of [research and development] and their sales models that really supports it. Subsequently, you've seen what were the largest server companies in the world lose money in their current enterprise businesses. The issue is when you move to standards, cost, logistics, distribution and efficiency become extremely important. Those companies are not set up to fly in that environment. Take Sun: They were always Unix and they're having trouble. HP is the largest Intel server company in the world, and even they don't make any money. So the real benefit is we already have a model that is refined for that standards-based, commodity-based environment. As we see more software and systems management software come and support standards - which we are - it flies into the sweet spot of our business model and leaves the sweet spot of our competitors' business models.
Are you concerned at all about Sun’s move into the low-end server market and HP’s plans to introduce a low-end server family? We look at them and say fine, have at it. This is a business of cost reduction we’ve been doing for 20 years, so we’re confident that our cost-reduction activities will be equal to or better than what they are going to do. I’m not too worried about Sun. This is such a massive change from Sun’s traditional strategy that anyone has to look at that skeptically. HP is a little different in that they have had a very large Intel-based presence, so they can certainly demonstrate that that’s not that far off their business model. But the problem is being able to make any money at it when you get into very low cost. HP has a distribution model which is very high cost and their whole reseller model never has been competitive with the direct model.
What do you think of high-level directions such as utility, autonomic, grid or on-demand computing? How real are they, and do they pose any threat to Dell's business? They are very fancy buzzwords for models that have existed in the past that have proven unsuccessful. If you assume that the status quo will never change and that companies will continue to have all the complexity and challenges in the data center, there will be some of them who will want an on-demand type of capability. The vast majority of systems that are sold are not to that type of environment. They are to small and midsize businesses and large companies that are not looking for that kind of value and are finding they can move to a grid structure, which is more aligned with standards and simplicity. They actually can get more cost savings, and they are not outsourcing or 'de-skilling' their company in an area that they, frankly, are going to need in the future. I don't think you can find any company that has outsourced its entire IT infrastructure in the on-demand mode and been a leader in their industry.
How does Dell view on-demand technology, and how are you meeting demand for that with your customers? We are meeting it with the standards, which say as you need more capacity you buy more thin, low-cost servers and you add capacity incrementally. The problem with on-demand is fundamentally you are buying a great big mainframe and paying for the portion you use. Someone eats the cost of that unused capacity somewhere. Basically, it's a strategy to sell big iron, because it is either going to cost the customer money or IBM money. We've seen companies go away from Sun systems because they had to buy huge capacities and fill it up over time. A much more efficient model is to buy just what you need and as you need more buy more thin servers with standard architectures that can handle all of your needs while you grow. The pay-as-you grow method is more efficient for suppliers and for customers.
But are there limitations? Are there applications that don’t scale out efficiently or can’t be run in a distributed environment? Sure, there are instances today where scale-out doesn’t operate. But if you go back five or 10 years, there were many, many more. We're finding more and more applications every day that are able to port over. The bigger challenge is not in the application porting, but in the middleware, or systems management, as you're managing many, many servers now. As you go to multi-core servers where you have many, many processors operating in a very thin server environment, the scalability is just going to get better and the cost of processing MIPS is going to get better, as well.
So how deep into the data center will Dell go with its servers and storage? There are two questions there: How deep will we go and how deep do we have to go? This has been a 10-year migration. We are very happy with the speed of the migration, but eventually we are going to go into almost all of the data center environment. We need some software help on it, clearly, but that’s coming. I don’t think there is any barrier to where we can go in the data center.
What markets will Dell look at next for commoditization? One we are looking at right now is the storage market in the enterprise. Further down the road, we’ll be more aggressive in the networking and switch market. Services overarching all of these has always been one we think will continue to commoditize, but it will do it in parallel with the hardware.
As far as the switching market, do you mean looking up more at the enterprise level? Yes. Today we are at Layer 2 and 3 in both managed and unmanaged. But over time, moving up into chassis-based systems will be likely, particularly if they go into some type of blade architectures. You have so many folks right now in that industry, it needs consolidation because they can’t all survive. When it consolidates and gets more simple and more standardized, as it will, that will be the time when it will be right for Dell to more aggressively pursue it.
What about services? Do customers look at Dell to provide services per se? They look at us for services that surround our hardware. They don’t look at us as an Accenture. We don’t have any intention of going there. We just intend to wrap services around the hardware that enable the hardware sale and enable the management and running of the hardware and are less concerned about the consulting piece.
What are the factors that you look for in markets in order to be able to enter them? There are several things. One is the emergence of essentially a merchant component industry. You've got to start finding the movement from proprietary ownership of all the product to a merchant ownership where you can see suppliers making the molecular components that go into the product, rather than one company owning it all. The second thing we look at is a kind of umbrella profit pool, meaning a large enough profit pool that is worth collapsing, but still be profitable. Third, we have to look and see customers are requesting or suggesting that Dell participate in the marketplace so we have some customer pull on what we're going after.
Looking ahead three to five years from now, how will Dell be different as you respond to things that you're seeing in the market? Well, we have a $60 billion target for the company, or goal I should say, that we're about two years into a five- or six-year program. So far we're on track. That shows we would have only about 50% or less dependency on client business, and the rest would be in enterprise products, servers and storage, networking products, services and then software and peripherals, which is to a great degree printers and displays. So that is what we had envisioned. We're wondering now whether this whole notion of the consumer electronics industry might be a little bigger, so that might skew it a little bit toward the consumer side. But we will still have the predominance of our products outside of the client, which are desktops and notebooks.
Are there some future technologies that you feel aren't getting enough attention? Bandwidth and wireless capability we wish were accelerating a little faster. There is another area in the standardization of systems management so that you have uniform management across storage, telecom and servers. That software technology capability really hasn't emerged in a standards form yet. That's something that could use a lot more pushing because we're going to need it for the myriad of clustered thin servers in the future. Customers don't want to have a multi-system management environment. They'd like to have one that could manage all the hardware and all the systems.
When you meet with customers or prospective customers, what do they ask you about most frequently? Well, it is manageability, probably. They are worried about security, too, but I think that's something that many companies are grappling with or dealing with anyway. I think it's manageability and cost. Those come up more often than technology.