Twenty-three years ago, the husband and wife team of Stanford University computer support staffers Len Bosack and Sandy Lerner founded the most powerful and valuable company in networking: Cisco Systems. This week, Bosack rolls out what he believes is another breakthrough product for enterprise networks that may make as much of an impact as Cisco routers. He shared some thoughts on his new product, the DXM optical transport system, and 16-year-old company, XKL LLC, with Managing Editor Jim Duffy.
Twenty-three years ago, the husband and wife team of Stanford University computer support staffers Len Bosack and Sandy Lerner founded the most powerful and valuable company in networking: Cisco Systems. This week, Bosack rolls out what he believes is another breakthrough product for enterprise networks that may make as much of an impact as Cisco routers. He shared some thoughts on his new product, the DXM optical transport system, and 16-year-old company, XKL LLC, with Network World Managing Editor Jim Duffy.
See also:Cisco founder brings optics to the router guys.
XKL was founded in 1991 to do computer I/O and backplane research. What made you shift gears to optical?
We actually constructed a complete computer system that was a much smaller form factor replacement for a then-obsolete product made by Digital Equipment, which was one of their mainframe systems. So essentially we produced a half-rack unit system that replaced something that was the better part of 20-feet long. We sold a few of those to some of the then remaining customers.
By 2003, it actually appeared as though the industry was ready to deliver on some of those promises (made during the bubble years of 1999-2002) in a way that we could see how to build a dense economical system that could be used by network builders as opposed to just telephone companies. So we took a lot of the work that we had been doing and tried to produce a really dense, potentially economical system. We were making a wager that the optical component makers could both deliver upon their promises and that they could actually do something that they were never very good at – make the price go down. And here we are almost four-and-a-half years later…
So you see a lot of press releases from people saying the bright, bright future is here and it’s all wonderful and it’s easy to do and “Gee, why aren’t we going from 40Gbps to 100Gbps because 40G is so yesterday?” That’s just not the truth.
So DXM is intended to bring optics to enterprise network builders with the familiarity of a Cisco router?
The people that we are probably going to sell the most of these devices to already have multiple campuses that already are buying multiple DS-3s. There are three prongs to the thinking about this product. One of them is an economic fact: A lot of fiber went into the ground during the bubble and (the people deploying it) didn’t really have a coherent vision as to what they were going to do with it. The financial results of that is that the current owners of these bits of glass in the ground can charge a pretty reasonable lease fee and feel that they’re getting an excellent return on their invested capital.
Another prong is, computers keep getting faster. And what we see is that at these speeds there’s no way in the world that humans are going to be a source of that. It’s computers that do it. So then comes the observation, well where are the computers? Corporations own computers and they have a need to have them talk, manage them, back them up… When there’s enough bandwidth around it changes how you can conduct your business.
So get all that hooked together and do it in a way that you’re not paying 45Mbps at a time every month. It turns out, because of the finances of these things, if you go to the trouble of running usually just across the parking lot to get to the fiber, you can get a phenomenal improvement in the cost per bit per second. It’s not uncommon to see the cost per bit per second went down by a factor of 100, or even 500, or actually a factor of 1,000. Conventional wisdom is, when you change the price or quantity of something by a factor of 10, it’s not just the same thing anymore, it becomes a kind of different quality. And when you do it by a factor of 100 or more, it’s an even more dramatic change.
So these are enablers for people to change the way they run their businesses internally. If you’ve a few thousand (computers) in a metro area, this is something that really does change how your business gets organized.
What price per bit per second are you quoting to your potential customers?
I don’t think we actually usually quote a particular price but here’s how it typically goes: A DS-3 is usually a couple thousand dollars a month, depending on where you are and where it’s going. For a fiber ring around a metro area, if you’re only stopping in three or four places you get a monthly recurring charge in the $20,000/month range. And there’s a one-time cost to get to the fiber rings, and it’s usually $20,000 per location. That’s when you actually have to dig a ditch to get there. There’s the cost of the equipment, which in our case is, for 100Gbps pizza box, is 1/8 of $1 million, basically. Four of those is $500,000 in equipment. You got yourself 200Gbps of bandwidth out of that… that gives you 10 cents per megabit per second.
You see how the numbers come out. Even if it were 10 times more expensive, it’s such a change that it’s almost astounding that people shouldn’t be running to do this because it changes what they do in their business. They stop arguing about a scarcity.
These are times of plenty. We are the bringers of inexpensive bandwidth.
Do you think the DXM has the potential to impact the market or the industry as your first Cisco router did?
If anybody had claimed in 1984 that you could purchase a graphically displayed GPS system powered by batteries for $99 or less, no one should have believed you. At the time you were spending $250,000 for two seven-foot racks and at best it read out the numbers on a display…
I can’t quite make predictions like that it’s going to change the whole world; it’s all going to be different. The best I know I can tell you is, when you’re getting bandwidth at 10 cents per megabit per second it’s really different than what you see now. Something’s going to change because of that. And I can’t directly tell you what but I am sure that something’s going to change.
Do you anticipate any resistance for early adoption?
In the early days of Cisco we’d occasionally get the question of, “Well gee, if this is good for my business why doesn’t the phone company want to sell it to me?” People who in the early days went and built their own level 3 networks found that they could change things about how their organization worked. One firm in the upper U.S. Midwest had design operations in Germany and components suppliers in East Asia. They were literally having buckets of things sent overnight from place to place – disks and tapes and drawings and things. The ability to have them speak in real time in any realistic way completely transformed the rate at which they could work.
You have a bit of that same first-mover advantage, to use an old, tired phrase here. If you go to the trouble of providing your own low-level transport because you bothered to dig the trench necessary to get to the fiber, you will have an advantage over people who do not. That old word “transformation” is used a lot in business and this is an actual change in the way you can use communications and computing in your organization. Those folks are going to be able to do things that people who are still paying by the DS-3 are not. Look at the change in cost – it’s not just the change in money, it’s the change in what you can do. Sure, there’s a capital chunk; but it’s not an enormous step to go to double that bandwidth, or triple it.
So yeah, this is a case where the first folks to do this compared to their competitors are going to be able to organize themselves and do things that their competitors can’t do. So they will have an advantage in their field.
What’s your impression of the industry now since founding Cisco, and of Cisco’s influence in it?
Cisco’s doing fine. They have to be all networking things to all people. As they’ve grown, they have a very different business problem today than the ones that I’m interested in. They need to literally be all things to all people, ranging from friendly little things that people who ought not to know much about networking can install in their homes, to the CRS-1. They have many, many specific things that they have to do one market at a time. That’s necessary for them to be the universal network provider.
It’s good to see the benefits of broad, efficient data communications being made available around the world. I just wish it had gone faster. There’s still access problems (that are) primarily political, not technical. Gosh, I wish that weren’t true. I wish there was more uniform, easy access the world round. But the business largely has developed very well. I’m a little surprised at the rate at which the money’s grown in terms of the size of the business. I had certainly expected by the late 1980s that it was going to be possible to produce $100 routers. But I’m not sure I would have guessed that there would have been $39.95 routers. And that they would be sold in such vast quantities. That the technology had a significant impact on the way the world works wasn’t surprising; yet again, I offer you the observation about GPS. Had I made the prediction then about what did happen, no one should have believed me.
