[Iccrg] Flow Rate Fairness: Dismantling a Religion
Bob Briscoe
rbriscoe at jungle.bt.co.uk
Tue Oct 17 00:03:20 BST 2006
ICCRG folks,
Here's the memo on deconstructing the Internet community's dominant
fairness ideology that I promised. It should appear in internet-drafts
shortly, but you can pull the tech report off my Web site now if you prefer.
To some of you it will not be anything new, but I have found that very many
people have got into a trap in their thinking, so I'm trying to jump people
out of what is actually a self-referential dogma. Concepts of fairness in
TCP, WFQ and so on are all afflicted with this dogma which is actually way
off beam relative to any reasonable views of fairness in real life. It's
deliberately blunt and simplified.
Flow Rate Fairness: Dismantling a Religion
<http://www.cs.ucl.ac.uk/staff/B.Briscoe/projects/2020comms/refb/fair_tr.pdf>
Abstract: We were moved to write this memo because the applied research and
standards communities in networking are using completely unrealistic and
impractical fairness criteria. The issue is not whether they should use
this or that allocation scheme; they don't even allocate the right thing
and they don't allocate it between the right entities. We explain as
bluntly as we can that sharing out flow rates (as TCP and many other
popular fairness mechanisms do) has no intellectual heritage from any
concept of fairness in philosophy or social science, or indeed real life.
Comparing and controlling flow rates alone will never achieve fairness and
should never again be claimed as a fairness mechanism for production
networks. Instead, a realistic fairness mechanism must share out the `cost'
of each users actions on others.
Here's the same text formatted as an I-D, but it becomes twice as long:
<http://www.cs.ucl.ac.uk/staff/B.Briscoe/pubs.html#rateFairDis>
I may not be able to get into discussions over the next week or so, given
other deadlines...
Bob
____________________________________________________________________________
Notice: This contribution is the personal view of the author and does not
necessarily reflect the technical nor commercial direction of BT plc.
____________________________________________________________________________
Bob Briscoe, 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