I sincerely congratulate you for Geoff Huston's excellent article in The Internet Protocol Journal, June 2008, on the "Decade of Internet Evolution." The article shows an amazing insight into the Internet as it has recently evolved and deserves as wide an audience as possible.
The only comment I could make is that though Huston hints about separating the IP address from the host name, he does not explicitly mention the Host Identity Protocol (HIP) . Previous issues of the Journal have this omission as well.
Note: As we struggle in the IETF and everywhere else in the industry with NAT traversal, mobility, and multihoming, we see countless approaches for each application layer protocol separately. HIP seems to fulfill the promise of solving these problems comprehensively.
Thanks for the privilege to continue reading the Journal; keep such papers coming.
—Henry Sinnreich, Adobe Systems, Inc.
The author responds:
Thank you for your generous comments.
At some point I was toying (dangerously!) with writing an article that attempted to predict the next 10 years, looking at what appears to be important today and what that could mean in the future. There is no doubt that the tight binding of identity and location is one of the assumptions that has made the Internet both simple and effective for the past decade. But where we sit today, in a world dominated by scale, mobility, a dense mesh of interconnectivity, highly capable end devices, dense middleware, and a panoply of specialized requirements, we need to look forward to methods that allow separation of identity and location. Now this separation could be at the level of the Internet Protocol itself, as in HIP or Site Multihoming by IPv6 Intermediation (SHIM6); or at the level of the transport session, as exemplified at present by the Stream Control Transmission Protocol (SCTP); or even at the application level, where the various offerings related to Voice over IP (VoIP) and Peer-to-Peer (P2P) have been working at the level of multiparty application rendezvous and application identity that sit on top of an adaptive platform of dynamic discovery of the characteristics of the underlying transport subsystem.
Each approach appears to offer some significant leverage in scaling the network in diverse ways, while at the same time presenting us with some fascinating insights into possible architectures that could address our needs in the next decade. No doubt the next 10 years will present us with some quite novel challenges with the imminent exhaustion of the unallocated IPv4 address pool and the associated observation that the schedule for the update of IPv6 has proceeded so slowly that we will be forced to be remarkably inventive with IPv4. HIP may well be a central part of such invention, but, more generally, I have no doubt that we will examine more generally how we can devise refinements to the networking model that preserve useful notions of identity across a rather fluid sea of shared location tokens.