WWW2010 – Vint Cerf Keynote

So WWW2010 is finally upon us here in Raleigh, NC.  As a Raleigite, it’s a little intimidating to see a rundown of past cities where this illustrious conference has been held, and then to see Raleigh in that mix.   Major props to the conference committee for bringing this to Raleigh.

Vint Cerf gave his keynote address this morning.  I thought I’d post some of his interesting observations.
Vint threw out some stats on Internet usage.  His late 2009 numbers indicated that there are about 1.3 billion PCs on the Internet and somewhere between 600 million and 800 million net-connected mobile devices.  There are 730 million visible hosts (those with direct IP addresses).

One of the biggest challenges facing the Internet is the fact that IPv4 addressing is running out — fast.  IANA will be out of addresses to give to Regional Internet Registries sometime in 2011.  Then sometime in 2012, the RIRs will be out of addresses.   Vint joked that “the Mayans must have known something”.

Vint said that it’s not necessary to fully cut over to IPv6.  Instead, you can add v6 on top of your existing v4.  I haven’t really given a lot of thought to it — it always seemed so far off.  But I think it’s time to talk to my hosting provider about their strategy.

Vint spoke about security on the network, and acknowledged that the early designers of IP networking didn’t really think much about security, as all the users were trusted.  But this is becoming more and more of an issue as attacks move into the realm of organized crime and even state-sponsored cyber warfare.

He gave a shout-out to stopbadware.org, which is working to rid the net of viruses, spyware, etc.  One point he made is that as organized criminals profit from botnets and the like, they have a vested interest in making the presence of their software as invisible as possible.  So the detection of such malware will only become more difficult.

Cloud computing was a topic of discussion.  I think the most interesting point that Vint made was that today’s cloud computing is similar to the state of computer networking in the early 1970s, when each vendor or research organization had their own fully-functioning, but isolated, networks.  There was no concept of communicating between these networks.  He threw out a challenge to the designers of today’s clouds to develop a baseline standard for data exchange and collaboration between clouds so that we can move our data and computation smoothly from one cloud to another.  There’s no need to standardize on a single cloud design/implementation, he said, because clouds are designed for so many different purposes, but they should at least have a common mode of interchange, much like telnet provided that baseline means of communication between early IP networks.

Vint spoke about the divide between our current storage capacity and the processing capabilities to handle the tremendous amounts of data that we’re able to create and store.  He suggested bringing the compute power and the storage together in powerful clusters.  His analogy was to take a rack full of servers, each with its own processor(s) and local storage, and then squeeze that down into a single subsystem with multiple processors and storage units working in concert to provide more efficient retrieval, processing, and analysis of massive datasets.

A thought-provoking topic was that of bit rot and inaccessible objects.  How would a user from the year 3000 use an .xls file from the year 1997, for example?  It is important for us to not only preserve our data objects, but also the means with which to access those objects.  This got me thinking about a field of “data archaeology“, which as it turns out, is a real field.

It was great to hear from one of the guys who started it all.  I can only imagine what he must think every day as he watches developers make use of the tremendous platform he helped provide.

Leave a Reply

Your email address will not be published. Required fields are marked *