This Woman Changed the Internet - PART 2
This Woman Changed the Internet - PART 2
In the second part of our interview with Sun Microsystems’ Distinguished Engineer, Dr Radia Perlman (PhD) Greg McNevin discusses a range of issues from Open Source to control of the Internet, and whether Ethernet was ever ‘cool’.
IDM: Does the fact that Ethernet is dominating mean that we will always at the mercy of it and its bridging devices?
RP: People should not be aware whether they are using Ethernet or fibre-channel or whatever. It’s not something that users should be aware of. This is also true for configuring web proxies, it should just work. So, we need to make networks a whole bunch more self configuring.
Now, as for Ethernet, one of the things I like to explain in talks is that a lot of what we know is false. One of the things we know about Ethernet is that it is very successful. But it’s not. It’s dead. The technology that the people who invented Ethernet were inventing, does not really exist anymore.
They invented this thing called ‘carrier sense multiple access with collision detect’. It was cool technology, but that’s not what Ethernet is today. Now where bridges came about, people were very confused by Ethernet. I applaud the people who invented it, but I’m annoyed at them for calling it a network. It’s not a network it’s a link. A multi access link. Back then IP was only this kind of researchy thing and the really big commercial thing was DECNet. But DECNet is sort of functionally equivalent to IP. So there was this big press excitement over which technology was going to win.
I was doing DECNet and I saw Ethernet and I though “Oh, I have to rethink the routing algorithm to make it efficient when you have links with hundreds of thousands of nodes”. So I changed DECNet to accommodate this new type of link. But people thought that Ethernet was a competitor to DECNet. Ethernet as it was originally envisioned did not scale beyond a certain distance. It couldn’t scale beyond hooking up a building or something. So there was a group at digital doing a protocol and they were doing it right on top of Ethernet and leaving out DECNet.
So, I went to them and I said please you really want to build on top of DECNet. And they told me “Oh, you’re just upset because your layer is a dinosaur and we don’t need it”. I said “You may want to talk from one Ethernet to another”, and they said,“No, our customers would never want to do that”. So they built their technology on Ethernet. They would have made as much money as if they made it on top of DECNet, because hey! our customers do want to talk from one Ethernet to another.
So, bridges were a way of figuring out how to let stations on one Ethernet talk to stations on another without them cooperating. To use a router you have to implement IP, or DECNet, or some Layer 3. Ethernet was not designed to be forwarded. It doesn’t have a hopcount, the addresses don’t reflect where you are in a topology, so it would be like the postal service trying to deliver a letter to addressed to just: Radia Perlman. How do you know where I am? I don’t know where I am most of the time.
IDM: ICAAN has highlighted a worldwide desire for control of the Internet to be less U.S.-centric. What are your thoughts on this?
RP: I’m not sure what it means to control the Internet anymore. The thing that hands out numbers for protocols, you have a field that has various values and someone has to hand out the values. People are fighting over who gets to do that and it doesn’t seem like that much of an important function. You could misuse it and refuse to give people numbers, but other than that I don’t think it matters much who does it. There’s the name space, where it really is unfortunate that you have .com.au and .com without anything at the end which means US. You know that sort of indicates some bias. It would be nice to re do the namespace so it’s not so US centric. It probably will never happen and it probably doesn’t matter a lot.
IDM: Who is the most influential force in networking?
RP: Well, the politically incorrect answer would be Microsoft because they are powerful enough to be able to do their own thing regardless of the standards bodies. Then there are the standards bodies that are really influential, even though it’s very political and it’s really various companies that have various monopolistic positions that tend to run those. Maybe it’s the spammers and virus writers who are the most influential because they are changing the way we think.
IDM: Finally, do you think open source is the way to go? Do you think the Open Source community is heading in the right direction for the rest of the community?
RP: Open Source would be wonderful. It’s the right thing that makes things more secure because people can evaluate it. It’s amazing that it has done as well as it has and I hope it does even better.
Dr Radia Perlman...
...is the inventor of the spanning tree algorithm. She has a PhD from MIT in computer science and specializes in network security protocols. This is the concluding part of our interview with Dr Perlman.
Comment on this story
Related Article: