[Iccrg] Proposal for ICCRG operations

[John Leslie]

Shivkumar Kalyanaraman <shivkuma at ecse.rpi.edu> wrote:
> 
> BTW, what are the directions are we really interested in for this group?

   This is a very good question. (I suspect different folks come with
different preferred directions.)

> - long BD products?

   I think we need to tackle the bandwidth-delay problem.

> - loss tolerance in wireless?

   I don't think we can tackle this directly until we decouple
congestion-awareness from packet loss (which I believe we should do).

> - more robustness to variance in underlying links?

   I don't know what the level of interest here is. Personally I'd like
to rate links on bandwidth and latency, with expected variance in each.

> - fairness considerations (max-min/processor sharing vs proportional etc)?

   This one is a bit scary. I'd prefer to defer fairness flamewars until
we've got some actual algorithms in use, being actively measured.

> Is there a priorization of relevant concerns? I only saw a broad wish list
> in the original talks...

   In research, traditional "prioritization" doesn't apply: work of interest
to individuals with time available gets done; other work doesn't.

--
John Leslie <john at jlc.net>