Friday some friends from work and I went to RIT‘s first Dean’s Lecture Series talk of the year by Peter Freeman. This talk, about GENI (Global Environment for Network Innovation), was informative but a lot less interesting than I imagined. I thought there would be discussion about their thoughts for the next internet and where we might be going. Instead we were inundated with boring generic statements of how GENI will be a testing bed for experiments dealing with the next internet. All of this is available at their website but to summarize, GENI’s primary objectives are:

• To develop and evaluate ideas for future network design
• To encourage related research

Some things that struck me during the talk:

1. The importance of a comprehensive coordinated effort in order to avoid the same defects existing in today’s solution was one of their key points. In almost in the same breath they mentioned the different approaches by Japan, The European Union and The United States.
2. Their basic architecture included devices named super routers. If these are like today’s routers they’re already building some very blatant similarities into what’s supposed to be a test bed for a new architecture. One of the things they mentioned was changing the TCP/IP stack. Routers are level 3 devices and as such currently utilize the TCP/IP stack. If these new devices are different than today’s routers they should have a different name.
3. It seemed to me they were concentrating purely on the hardware networking part of things. Isn’t our hardware pretty solid? Can’t we already have 5×10²⁸ addresses for each of the estimated 6.5 billion people alive today with IPv6 (wikipedia)? Can’t we already handle that bandwidth with the existing broadband technology? I would think the bigger concern is archaic protocols such as FTP and HTTP and their underlying stacks like TCP/IP, which have been hacked together over the years. DZone recently posted an article about why FTP Must Die and its definitely worth a read.
4. If the hardware is redefined but the software and protocols are not, won’t we just end up with one giant hack which fits all of today’s technology into tomorrows architecture? This just seems like adding an extra layer to me.

GENI has a good idea with their general principle. We do need to be looking to the future, some of our current internet practices simply wont be able to hold up when

Every Light switch has an IP address.

However; they’re going down the wrong path and it seems like they’re creating too much (hardware) infrastructure to really inspire the creation of a new and unique solution.