Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

[RPRWG] Fairness work items





Folks,

Here's a list of the work items that came out of the 
Fairness Section comment resolution.  

If you see your name listed against something and if 
you don't like it, please let Necdet or me know as soon 
as possible so that we can look for possible alternatives.  
(Hopefully this applies only to those that weren't present 
when they were volunteered for the effort by the comment
resolution group!)

Feel free to ask for clarifications on any of the 
work items.  You can also refer to the CRD files to
get a better context of where these came from.

-Anoop
--
Anoop Ghanwani - Lantern Communications - 408-521-6707

===============================================================

- Peter Jones to come up with better names for Type A and 
  Type B messages to address his comment.  He will discuss
  the names on the reflector to gauge consensus.

- Necdet will generate 3 state machines - comments are welcome.  
  Others may contact Necdet to work on this.  The state machines 
  will cover:
  1. Congestion detection state machine;
  2. Fairness message generation;
  3. Behavior on receipt of fairness messages.
  They will be submitted as a contribution for the next meeting.
  The congestion detection state machine will address the issues 
  from this comment by indicating how hysterisis is provided to 
  avoid oscillations.

- Behavior of fairness messages during wrapping to be
  handled by Necdet's state machines (see above).

- Stein's comment on the description of the algorithm
  should be covered in Necdet's state machines.

- John Lemon to come up with a formula for calculating the
  convergence time if he cares about it.  (His comment
  was to remove the reference to simulation and 
  replace with a formula.)

- Clause 5 editors to remove 5.8.3.2 and 5.8.3.3 as these
  are already covered in Clause 9.

- Harry and Anoop will work together on a state machine/behavioral 
  description for a "conservative" mode of operation which allows 
  for other methods for computing the fair rate.

- Editors to come up with proposed draft text for a seperate
  section describing all the variables used by the fairness 
  clause.  This text to be posted as a contribution for WG 
  review at least 4 days before the close of the comment 
  period.  Editors to submit a comment for including the 
  text in the draft.

- Fairness frame format problems:
  To be forwarded to the RAH for detailed bit assignments
  with the intent that the Type A fairness frame not 
  exceed 16 bytes in size.

- Editor to coordinate with Clause 5 on the TTL issue.

- Sizing of STQ:
  Vasan Karighattam be requested to come up with a 
  contribution for this.  This will be submitted to the 
  working group at least 4 days prior to close of the comment 
  period for consideration for inclusion in the draft.  David 
  James, Mete Yilmaz, and John Lemon will assist by providing 
  early review of this contribution.  A comment should be 
  submitted for including the text in the next draft.

- Algorithm needed for "advertised rate" calculation.
  Editors to come up with proposed text to address this 
  comment and submit it to the working group for review at 
  least 4 days before the close of the comment period.  
  Editors will submit a comment to include the said proposed 
  text in the next revision of the draft.
  (Unfortunately, I'm not sure I fully understand what 
  this last one does, but I'll try to get a clarification
  from others.  In the worst case, it'll hopefully be 
  addressed by Necdet's state machines.)