[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Diffserv] Model - queues



Dan,

> > In general, you can have a buffer management scheme that continuously
> > makes state-dependent decisions on packets already in the buffer, i.e.,
> > you have the option of changing your mind even after accepting the
> > packet.  One of the simplest example of such a scheme that has been used
> > (with limited success though) in ATM switches is the pushout scheme, but
> > there are other variations that are possible and I think we should try
> > to have a structure capable of accommodating this.
> >
> Hmm.  I hadn't heard of this.  Do you have a reference?

Here are a few pointers, mostly papers analyzing performance, although I
recall at least one paper by Landsberg and Zukowski (Columbia U.) which
was describing a VLSI design they had done for a scheme that was
actually
more generic than simple push-out.  I cannot put my hands on the exact 
ref right now, but maybe someone else will recall what I'm talking
about.
Most of it is also described in P. Landsberg PhD thesis entitled "VLSI
for packet switching nodes in communications networks." (don't have a
copy of that either, but maybe Paul has an online version somewhere).

- H. Kroner, G. Hebuterne, P. Boyer, and A. Gravey. "Priority management
in ATM switching nodes." IEEE Trans. Commun., vol. COM-9, No. 4, April
1991, pp. 418-427.

- J. Garcia and O. Casals. "Space priority mechanisms with bursty
traffic." 
IFIP Trans. C. Commun. Sys., Vol. C-4, 1992, pp. 393-412.
> > So I would say that a discarding element is an element which for every

- J. Garcia and O. Casals. "Stochastic models of space priority
mechanisms with 
Markovian arrival process."  Anals of Operations research, Vol. 35,
1992,
pp. 271-296.

> > So I would say that a discarding element is an element which for every

> > state change in the buffer, i.e., a packet leaving or arriving, is
> > making decisions on which packets to keep/accept in the buffer based on
> > a discarding discipline.  This is not much different from what you have,
> > but I thought it might be important to clarify things.
> >
> What you're pointing out is that the description doesn't capture the notion of
> a trigger, which could either be fired by a packet arrival at the discarding
> element, some state change in the FIFO element, scheduler, or meter, or a
> timer.  That's useful.

And also that the discarding decision can apply to any packet in the
buffer,
not necessarily the arriving one or the one at the tail/end of a linked
list.
You could pick a random one (not sure why, though ;-) or actually more
than
one, e.g., kicking out several small packets to make room for a big one.

Roch
begin:vcard 
n:Guerin;Roch
tel;work:215 898-9351
x-mozilla-html:FALSE
org:UPenn;Elec. Eng.
version:2.1
email;internet:guerin@ee.upenn.edu
title:Prof.
adr;quoted-printable:;;University of Pennsylvania=0D=0ADept. Elec. Eng., Rm. 367 GRW=0D=0A200 South 33rd Street;Philadelphia;PA;19104;USA
x-mozilla-cpt:;0
fn:Guerin, Roch
end:vcard