While Compaq currently rules the server market the vendor is facing a growing threat from Dell, which is led by one of the industry's best-known entrepreneurs, Michael Dell. In this interview with IDG's John Gallant and Marc Songini, the eponymous Dell discusses the differences in strategy between his namesake company and Compaq, Dell's future plans for the server market and the role of NT in enterprise applications. Dell also explains why he doesn't like the idea of thin clients and talks about why his company won't be following Compaq's lead in acquiring network companiesIDG: What do you see as the key differences in your server strategy compared with companies such as, say, Compaq or IBM?
Dell: We changed the rules of the game in the server market by being the strongest proponent of the move towards an open architecture and by taking out some of the proprietary margin that existed in the server models. We redefined what the customer was able to get at various points along the curve of price/performance.
While most of our effort has been in the entry-level and mid-range of the NT server space, in the second half of this year we are going to move from being a fast follower in technology to technology leadership with a focus on greater rack density, storage, clustering and performance. This is going to change what customers have to pay for this technology.
How far do you think NT and the Intel architecture can push up into the enterprise?
I don't have delusions of NT replacing all server platforms in a year or two. However, we're in a progression that is going to go on for the next five to seven years. As we get the Intel/HP Merced chip at the 64-bit address space, the opportunity to replace the high end of the minicomputer market and even some of the low end of the traditional mainframe space certainly exists.
One of the other differences that highlights Dell in the server world is we have a significant amount of our attention applied to NT and are not putting a great deal of energy into other platforms. Our strategy is to go where thegame is going to be. We see NT as the primary growth engine. While it's not meeting everyone's needs, NT is where the vast majority of the growth is.
Does the Compaq/Digital deal change your server strategy at all?
No, it doesn't. The questions we get around this relate to services. We have aligned with both the leading maintenance and break/fix companies - Wang, Unisys and so on- and the leading outsourcing companies like EDS and Andersen Consulting. Our focus is to offer the customer choices.
We don't see that bundling products and service is the right answer.
What's the secret of Dell's success in the server market?
We sell direct. This is a powerful economic force in computing. It's changing the way customers are buying. The primary benefit to the customer is that you are eliminating the reseller markup. The fact is that in new technology the cost of materials declines about 1 per cent per week. So if a competitor has 70 or 80 days of inventory and we have seven days of inventory, that's about 10 times more inventory, and a significant difference in the cost of that to the customer.
There are a number of customers who are concerned about this dealer markup and inventory disadvantage.
Compaq has bought a number of network vendors and has tried to make the case for buying both servers and network components from one vendor. Do you see that as a valid strategy?
There are components of networking that are tied to the server. The obvious ones are the things that go in the box - the network adapters and that sort of thing. You could make a case that things attached to the server could be easily sold. But Compaq has to convince the reseller to sell the Compaq server and the Compaq router or the Compaq switch, rather than the Cisco or 3Com or Bay Networks product. That's not easy. Larger network users have bought into architectures. For somebody to switch them, there's going to have to be a pretty compelling argument presented. I'm not sure it has been.
Can you clarify Dell's stance on thin clients?
We see the network computer as counter to the whole move to Windows NT. So for the NC to take off, the trend toward NT has to stop and reverse itself, which we don't see happening. The other problem with the NC is it doesn't have any support in an unconnected fashion. So if I take those users that are going to NT and those users who require mobile computing out of the equation, we're talking about a very small portion of the market.
Is there a thin-client model that you think is more interesting? Say, the Windows-based terminal model?
What you're likely to see emerge is the best of both worlds - the managed client that takes away the users' ability to cause events that add cost to a company. The state of that machine would be stored on the server, so you've got a robust and redundant capability. But we're not promoting the NC architecture.
But do you promote the Windows-based terminal architecture? Would you offer products in that market?
Not particularly. It tends to have a fairly narrow appeal in medium-sized businesses.
What should network managers expect in a Merced/NT 5.0 world?
The first thing is they shouldn't expect it anytime soon.
But we've got an opportunity to take this whole open microprocessor computing architecture all the way through the minicomputer and up to the low-end of the mainframe architecture, and do it with an economic model that is more favourable for customers and provides more choice.