Bay Networks president and CEO Andy Ludwick recently outlined his company's vision for networking hardware in the coming age of intranets. IDG's Stephen Lawson asked Ludwick how the intranet revolution will affect network hardware architectures.
IDG: Can you explain what you mean by intranet switching?
LUDWICK: The Intranet Switch is a combination of a hardware platform and software. The hardware platform is Layer 2 switching, so we get the advantages of high bandwidth to the desktop, near the desktop. The software is routing. The Layer 2/Layer 3 switch allows routing decisions to be made in a decentralised architecture, close to the edge, and close to the desktop. You're getting the benefits of high bandwidth and decentralised routing solutions in one integrated platform. It's a platform for combining routing and switching, pushed out to the edge.
IDG: How does an enterprise migrate to an Intranet Switch platform?
LUDWICK: They use the same architecture as today, where there is a campus enterprise backbone. They also maintain their high-end routers where they are as WAN interfaces, and then deploy the Intranet Switch in the wiring closet to provide this interface close to the desktop. So they use the same wiring closet infrastructure and the same backbone architecture, and they keep their routers in the centre of the network - but then just surround them with these Layer 2/Layer 3 switches out close to the edge in the wiring closets. Today what we have is a Layer 2 switch out on the edge and a router in the centre. We're integrating this. This isn't two boxes; this is an integrated technology. IPNNI [Integrated Private Network-to-Network Interface] is the key to the integration. IPNNI integrates Layer 2 routing decisions and Layer 3 routing decisions. They become one routing decision, made near the desktop sitting on a switch platform.
IDG: What's your remote-access strategy?
LUDWICK: To combine on the enterprise end and the access end - an integrated technology and an integrated network management system. Today, those are done separately. On the access end, people worry about simplicity and low cost - that's good. On the enterprise side, they worry about network management and serviceability. What we're doing is combining the two. We've been building and are now marketing a system that has the scale on the enterprise side and network management control of each end point; and then the simplicity and ease of use and low cost on the other side. We have what we're calling a universal access strategy that ties both perspectives: the CIO's concern for control and manageability with the user's concern for simplicity and ease of use.
IDG: A few years ago, the networking hardware business was in chaos. Now it seems like technologies such as Fast Ethernet and Ethernet switching have fallen into place to some extent. What happened?
LUDWICK: A lot of the questions that were around a year ago have now been answered. A lot of that has to do with the technologies coming to market. When a new technology comes to market where it's going to displace an old one or change it, like Fast Ethernet, there are these expectations that get set ahead of the reality of the technology. There's this period of confusion, sometimes chaos. Sometimes customers will hold their business and wait. Then the product comes out, and you can see what it really does, and the application architecture starts to resolve itself. What we're seeing now is that there are certain desktop technologies: There's segmented shared; there's Fast Ethernet, meaning 100 shared; and then there's switched. Those all become nicely positioned for the desktop. We then have the edge technologies, where we're seeing frame switching. We'll see later this Layer 3 switch on the edge. Then we're seeing the core of the network, where routing and ATM are starting to evolve. The network centre, the edge, and the desktop have become clear elements of the customer's architecture, and technologies have started to show up and be real or not real for each of those areas. The customer has a better practical view of what products they can use and different elements, different architectural elements of their network today.
IDG: Where does Gigabit Ethernet fit in?
LUDWICK: It's a possible solution to scalable bandwidth in the centre of the network and a competitive technology to ATM. It certainly is something that I think will have some reasonable share of the business. Whether or not it can do all the things that are needed is not quite clear yet. Here we are again with technology that is not here, and so we're looking ahead and sort of speculating what it can do as an industry. On the other hand, it will certainly solve some portion of the problem. Gigabit Ethernet and ATM are going to coexist in the marketplace; there's going to be a certain class of applications and customers who really want to get on the ATM path. There's going to be another significant part of the marketplace that will stay with frame technology as long as they can. Both will be out there.
IDG: How are applications on intranets going to affect the decisions that CIOs have to make?
LUDWICK: They have to have an architecture that is bulletproof, highly reliable, and highly scalable. One of the real challenges IS managers face is that when they go and write a cheque, they know that the growth curve for use on the system is going to be very steep. If they commit themselves far ahead of the growth curve, then they're spending money they don't need to spend yet, and they'll have trouble. If they commit themselves to something that can't keep up, they'll have to throw it away and get another system. So scalability, functionality, bandwidth, and performance are going to be key. The other element is control: They're going to be very concerned with network management and their ability to deliver reliability and performance at a cost. The industry can't keep up with the demand for technical skills to service the network. You have to somehow leverage network management technology and have a skill set that is somewhat centralised across the network. I think the CIO is really thinking in terms of architecture these days, more than just solving today's problems - he's really thinking about: 'How do I scale, how do I look ahead, how do I stay under control?'