1 / 5

Updates on LEDBAT Congestion Control Specification

The latest draft for LEBDAT's congestion control mechanism reflects significant revisions. The document now clearly separates sections for sender and receiver, discusses parameterization and fairness, and includes a new appendix for clock skew correction. Key changes include adjusting the delay target to 100ms, specifying standard TCP parameters, and maintaining random reshuffling for fairness. Editorial suggestions include the removal of applicability to TCP and discussions around the design's foundations, particularly in relation to TCP Vegas. This version aims to align with deployed mechanisms.

adamma
Download Presentation

Updates on LEDBAT Congestion Control Specification

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. draft-ietf-ledbat-congestion-03

  2. Editors/Doc Structure • New active editors assigned • J. Iyengar, MirjaKuehlewind • Document restructured • Ledbat CC sections grouped under sender/receiver specific • Separate section around parameterization and fairness • Clock skew correction discussion moved to appendix • From version 3 we also intend to remove everything beyond A.2.1 which is the deployed mechanism presently, everything else is speculative at best

  3. Editorial • Intro should include discussion on TCP Vegas which forms the basis for a lot of observations in this design • Applicability to TCP seems far out and we would like to remove it • Adapting TS to start measuring one-way delay is non-trivial • There should be some discussion around BW and the kind of links where Ledbat is appropriate • If somebody tries to run this on 10G for instance it clearly wont work as designed • Replace TCP with Standard TCP

  4. Parameters • Delay target • Delay target moved to 100ms from 25ms in Version 2 • Discussion on specifying a range, SHOULD be x but MUST be no more than Y • Version 3: TARGET parameter MUST be set to 100 milliseconds (matches implementation) • Gain • Reverting to the text from version 1, version 3 will say “GAIN MUST be set so that the max ramp up rate is the same as for TCP”

  5. Random reshuffling for fairness • We are keeping this even though parameter SHOULD be set to zero in the common case. Ok? • random_input() # random() is a PRNG between 0.0 and 1.0 # RANDOMNESS_AMOUNT is normally 0 RANDOMNESS_AMOUNT * TARGET * ((random() - 0.5)*2)

More Related