Rethinking the Internet Architecture

Regular readers of my posts will note that I have often spoke of the need for Enterprise and Web 2.0 architects to pay attention to the underlying network that acts as the basis for their applications. This may seem strange, after all television producers don't spend much time worrying about cable or electrical networks, so why should application developers worry about the Internet? But how will Cloud and Web 2.0 services thrive if the underlying network is increasingly incapable of meeting growing demand for Internet services?

The reality is that the Internet is breaking under several different strains. IP addresses, using the current IPv4 scheme, are running out with less than two years supply remaining according to many estimates. At the same time, the growth in multihoming coupled with further disaggregation of the IPv4 address space are putting a growing strain on Internet routing tables, meaning growing potential for Internet sites to become unreachable as routers struggle to maintain forwarding tables. Many believe that IPv6 is the answer to these problems, but while IPv6 addresses addressing shortages, it actually exacerbates the route table scalability crisis. (Nemertes released a freely available report last year on these issues).

This week in Boston marks the latest attempt to address the Internet architecture crisis as the Pouzin Society meets to launch a new effort to create an architecture that allows the Internet to scale to meet present and future demand. Those responsible for delivering Internet-based applications would be wise to keep abreast of these efforts.