“I Believe in Making Things More User-Friendly More Than I Believe in Training People”

12/06/2013

“I Believe in Making Things More User-Friendly More Than I Believe in Training People”

Radia Perlman, American networking and security protocol expert, Fellow at Intel Labs and regarded by the community as the “Mother of the Internet,” was keynote speaker at the Network Security Forum that was part of the LACNIC 19 event. Before her presentation, she spoke to LACNIC News, among other things, about her point of view on migrating to IPv6.

By Pablo Izmirlian

The Internet: Myths, Missteps, and Mysteries – this was the title of the presentation by Radia Perlman, the American networking and security protocol expert who participated as keynote speaker at the Network Security Forum (LACSEC) held on Wednesday 8th May during the LACNIC 19 event in Medellin She was also invited to speak at the Women and IT luncheon, where she discussed her experience in an industry where women continue to be a minority.

Perlman is known as the “Mother of the Internet” for her contributions, but this is a title she doesn’t like. A graduate of MIT and author of books considered Internet bibles throughout the technical Internet community, Perlman is currently a fellow at Intel Labs, the research department of the well-known microprocessor manufacturer.

The migration to the new version of the IP protocol was one of the central topics she discussed during her interview with LANIC News. “What’s really sad is not that moving to IPv6 is difficult today, but the fact that the solution is not technically as good as the one available in 1992,” said Perlman in reference to the protocol proposed by the International Standards Organization in 1992 and which was discarded by the Internet community. “The only reason it was not adopted is because standardization agencies compete with one another and don’t want to admit that what another agency does is valuable,” Perlman added. “I can’t imagine how much money the industry has wasted by not adopting 20-byte addresses back in 1992.”

What projects are you currently working on?

I work at Intel, the problem is that I can’t talk about all the things I do. Generally speaking, however, I work on network and security protocols. I believe in making things more user-friendly even more than I believe in training people. We create systems that work well for people. I try to really understand what the problems are and to find ways to solve them without telling people they need to get rid of everything they have and start from scratch with something new.

Is that possible?

Yes. Many years ago I created the Ethernet spanning tree, which is how Ethernet works today. The CSMA/CD that was originally invented no longer exists, now everything is about links between two bridges – and bridges use the spanning tree. That was good because it was largely self-configuring, you plug it in and it works, but it does not make the best use of bandwidth because in a tree there is  only one way to go from one place to another, so some links remain inactive. In fact, the reason we created the spanning tree at the time was because people were confused and thought Ethernet was a protocol just like the IP protocol, which could be used for the entire network, when it was actually intended for a network within a single building. Spanning tree was the way to expand Ethernet, but its goal was not to replace the Internet Protocol. Now that everyone has network stacks running IP, why do we continue to move Ethernet packets? The reason is that IP is not a very good protocol.

One of this event’s main topics is the migration from IP protocol version 4 to version 6. Do you think we’ll see a new migration in the future, from IP to something else?

Well, it’s like English: it really is a fuzzy, ambiguous language. You could invent a much better language –in fact such language has already been invented–, but as long as you can say everything you need to say in English, no one has the motivation to change. This actually applies to all languages. A network protocol is no big deal. You’re just putting something in an envelope with an address that says where it needs to go, so there’s nothing too different that can be done. Today, “software defined networking” is all the rage. This doesn’t make sense. Many people say, ‘We don’t know what it is, but this is going to change the world.’ There is so much talk about this that companies feel they don’t want to miss the boat, when in fact there is nothing to miss. I mean, the big mistake was that in 1992 we didn’t adopt the format proposed by the International Standards Organization (ISO). This format used 20-byte addresses, while IPv6 uses 16-byte addresses. It would have been easy to migrate in 1992, as the Internet was then a tiny little thing used mainly for research. Now it is so large and so critical that we need to invent things to make IPv4 last a little longer… It’s like a frog in boiling water. I’ve never tried this – and I don’t recommend that you do–, but supposedly if you put a frog in cold water it will simply stay there. If you heat the water slowly, the frog never finds the water hot enough to jump out of the pot until, in the end, it dies.

As a security expert, what would you say are the main concerns relating to Internet security?

There’s the issue of user authentication, which is just really bad. Usually one must choose between making things more secure and less usable or making them less secure but more usable. However, we have managed to come up with authentication systems that are neither secure nor usable.

 

We should be smarter than that.

All these user names and passwords, all these sites with their own rules about what a password should look like… Even if you have some kind of algorithm, you can’t use it because every site has different rules. It’s very, very complicated. Attempts have been made to create better things. It would be nice if each user had a smart card, I don’t understand why that’s so difficult (laughs). There’s the problem that you might lose it, but then again everything has a solution.

Subscribe
Notify of

0 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments