[e2e] the evolution of deployability

Dave Crocker dcrocker at brandenburg.com
Mon Dec 16 09:14:15 PST 2002


Tuesday, November 19, 2002, 6:19:43 AM, you wrote:
Jon>  - i wonder (and this was something that Mostafa
Jon> Ammar, one of the panelists at ICNP talked about very nicely) if one
Jon> could actually go so far as to form a proper (econometric, or systems
Jon> science or perhaps even ecological) theory of network evolution? we
Jon> must have other examples that someone could give us Bell, Hooke and
Jon> Chandle on ?

Jon> measuing complexity is a computing/information science capability, so
Jon> then what other things to we need to draw on?


1. In developing changes, we now have more people, with less experience,
participating in the standards invention process. There seems to be a pretty
good case for claiming that much of the work now being produced tends
towards the "big system" design approach, rather than the more narrow,
incremental "surgical" approach that used to be common in the IETF. Big
system designs are notoriously difficult to deploy and even harder to change.

2.  Besides larger scale simply meaning more nodes/components, it means more
administrations and administrators.  And they, too, have a wider range of
skills.  (Part of the mark of a maturing field is that there are more people
in it who are not deep "experts".  However this also means that their work
is less flexible, since their work is more constrained and regular.  This
creates change difficulty.)

These issues interact.  The current IPv6 is a result of claiming that we
could not make incremental changes to the IP infrastructure, so we had to do
a single, massive change.  However a massive change is exactly what is
difficult to do in a deployed system.

This suggests looking at a number of human-related research areas, both with
respect to barriers within individuals and barriers within groups.

 Dave <mailto:dcrocker at brandenburg.com>
 Brandenburg InternetWorking <http://www.brandenburg.com>
 t +1.408.246.8253; f +1.408.850.1850

More information about the end2end-interest mailing list