[Iccrg] Meeting agenda (really! :-) )
Bob Briscoe
rbriscoe at jungle.bt.co.uk
Mon Oct 16 23:03:44 BST 2006
Keshav,
I fully understood your thesis.
If a bottleneck had capacity to be able to allow some users to get a lot
more capacity, wouldn't it be uncongested? As we are talking about
congestion control, I assumed we were considering a scenario where traffic
demand is in an epoch that congests the resource (or risks congesting the
resource).
If it's not congested, we certainly do move into a realm where anyone can
have any rate and it doesn't alter fairness, because fairness is measured
by how much volume of congestion each user causes. (My next mail announces
the posting I promised which justifies this latter assertion.)
But, for me, the whole point of congestion control, and iccrg in
particular, is to ensure large amounts of capacity are highly utilised
while still ensuring low loss, low congestion delay and fair allocations.
I imagine that some/many hi-capacity networks will be burstily congested
but often there just won't be sufficient demand to fill capacity. However,
every time demand happens to all arrive at once, congestion control has to
work in all the 4 dimensions just mentioned.
Bob
At 14:51 12/10/2006, S. Keshav wrote:
>On Oct 12, 2006, at 5:00 AM, Bob Briscoe wrote:
>Bob,
>
>>Keshav,
>>
>>At 20:41 11/10/2006, S. Keshav wrote:
>>>Bob,
>>>This sounds like a good discussion to have. The need for fairness has
>>>been grossly exaggerated, in my opinion, and is hard to justify. If
>>>scheme A is fair to all users, and scheme B is unfair, but every user
>>>with scheme B gets more than the same user with scheme A, would you
>>>prefer B to A? I would.
>>
>>Of course. But such a situation would be extremely unusual. All the
>>practical scenarios involve conflicts of interests, with some users
>>getting more and some less.
>I guess this assumes that you are already at the Pareto frontier. As
>I discussed in my earlier email, it is not clear to me that this is
>necessarily the case with today's transport protocols (especially
>when one of the hops is on a wireless link). A hypothetical unfair
>scheme that improves every flow's performance would certainly be
>better than wasting resources.
>
>>I have to disagree that the need for fairness is hard to justify.
>>In particular, in a non-co-operative environment (like the
>>Internet, which is after all the object of debate), if the view of
>>fairness that prevails cannot be /enforced/, then some users can
>>take orders of magnitude more than their 'fair' share. Then, what
>>is the point of even having a congestion control protocol?
>Let me restate my thesis:
>Consider congestion control scheme A that guarantees fair allocations
>to all flows. Suppose I come up with scheme B that is not fair, but
>every flow that got an allocation r with A gets >= r with B. Some
>flows may indeed get *much* more than others, but every flow does
>better. I argue that scheme B is better than scheme A. Or, more
>obliquely, capitalism trumps socialism :-)
>
>Now, your objection is that fairness needs to be enforced. But what I
>propose is to have the fair share be the lower bound on the
>allocation for all flows. That is the point of the congestion control
>scheme. (For that matter, if only a majority of flows improve,
>perhaps scheme B is still better.)
>
>>>
>>>--Paraphrased requirements (eliding application level
>>>requirements) --
>
>BTW - this is not my wording - I just cut and paste if from the
>survey that Michael pointed to. I put it out as fodder for discussion
>
>regards
>keshav
____________________________________________________________________________
Bob Briscoe, <bob.briscoe at bt.com> Networks Research Centre, BT Research
B54/77 Adastral Park,Martlesham Heath,Ipswich,IP5 3RE,UK. +44 1473 645196
More information about the Iccrg
mailing list