Review of "Internet Indirection Infrastructure"

From: Tyler Robison (trobison@cs.washington.edu)
Date: Sun Nov 14 2004 - 23:56:35 PST

  • Next message: Alan L. Liu: "Review of Internet Indirection Infrastructure"

      This paper proposes i3, an overlay network that provides a general layer
    of indirection, which could be used for multicast, for mobility (a mobile
    computer could maintain stable connections despite the fact that its
    address may change), and other purposes. The core idea is that senders
    send data to the i3 network along with an id, and this data will then be
    sent to receivers who are registered with that id. Similar to IP
    multicast, but it does differ; the main way in which it differs is that
    instead of a 'join', i3 uses a 'trigger' (can specify a destination,
    including a stack of identifiers). One of the results of this is that the
    receiver can actually specify routes using these triggers, perhaps with
    each stop on the route acting as stages in a pipeline. This possibility
    alone is probably the most interesting part of the paper, as it is a very
    simple and natural way for the receiver to run the data through a number
    of stages (among other uses).
      I liked this paper, and thought it was pretty well done. Halfway
    through
    it I had a list of security flaws of their system in mind, but by the end
    of the paper they had addressed each of issues, and several more in
    addition. Specifically, there seems to be a lot of potential for creating
    triggers for other hosts, or for using the identifier stack to create a
    loop in i3, but both are considered in the paper, and the techniques they
    suggest for handling these sound like they'd be effective. They also
    tackle issues such as robustness and scalability, and its nice to see a
    paper that takes these into account, even if their coverage is somewhat
    lacking.
      One point that bothers me is that this could be fairly inefficient at
    times; they discuss the case where the sender and receiver are close by,
    but the i3 server is halfway around the world, and suggest resolving this
    by locating nearby servers, but unless this were very widely deployed
    there wouldn't be nearby servers in many, or even most, cases. But
    this is a fairly small point; overall, the paper was very solid.

  • Next message: Alan L. Liu: "Review of Internet Indirection Infrastructure"

    This archive was generated by hypermail 2.1.6 : Sun Nov 14 2004 - 23:56:35 PST