re: random responses

Eric Hoffman (hoffman@cs.washington.edu)
Thu, 11 Jun 1998 12:20:37 -0700 (PDT)

The history was reversed. Digital AN2's per-flow credit scheme
was designed in late 90 before ATM had any congestion control or
even circuit setup standard (these appeared in 93 or so); the original
AN2 had 128 byte cells, a link speed that wasn't a 4^k multiple of 56Kbps,
etc.

sorry, I thought you had mentioned that the AN1 had been rate-based.

one of the primary reasons given for not selecting credits was the
associated switch costs. however, it turned out that the end-system
rate implementations had different behaviours, and no one got around
to implementing reliably correct ABR machinery on the switch-side
(which was hardly cheap either)

[as a side node I once spent an entire day trying to track down a
strange error with our software only to discover that the dec
gigaswitch-atm was mournfully sending out credits on a vc we
had thought we could use]

let me ask a different question, do you think there is likely to be
large differences in areas such a stability and performance between
rate and buffer-reservation based credits?