Opened 10 years ago

Closed 10 years ago

#215 closed editorial (fixed)

editorial issues around Congestion Control

Reported by: cabo@… Owned by: cabo@…
Priority: major Milestone: post-WGLC-1
Component: coap Version: coap-09
Severity: In WG Last Call Keywords:
Cc:

Description

Salvatore Loreto notes (http://www.ietf.org/mail-archive/web/core/current/msg03042.html):

The section 4.6. Congestion Control needs some editorial work as at moment it talks at the same time of the network congestion control issue and of a mechanism to avoid to overloading a server (suggesting for the latter a similar mechanism to the one used by HTTP).

->

The section is indeed entirely about congestion control and not about server load control at all.
It may need to be improved to reduce the potential for misunderstanding here.

Cullen Jennings (msg03072s, t, u,v) makes a couple of technical points (e.g., he argues that CoAP needs to be more strict about the number of concurrent requests than HTTP is). These are technical issue that need further discussion.

Also, Salvatore notes (msg03039) that the reference to draft-eggert-core-congestion-control in 4.5 would point to an expired draft. As 4.5 now has expanded discussion of multicast congestion control, maybe the reference is no longer necessary. It could also be replaced by a copy of the first two paragraphs in section 3.4 of draft-eggert-core-congestion-control (the third paragraph is now covered by material in 4.5 that was not available at the time 3.4 was written).

Change History (4)

comment:1 Changed 10 years ago by cabo@…

Also: - Comments

Cullen Jennings notes (msg03072l):

Section 4

Clarify that the stop and wait is per flow. Or is it per destination? Given the later suggestion that every transaction could be on a new port, there is a big difference between the two.

-> per flow.

comment:2 Changed 10 years ago by zach@…

  • Owner changed from draft-ietf-core-coap@… to cabo@…

comment:3 Changed 10 years ago by zach@…

IETF Vancouver WG meeting had a productive discussion including transport area folks. The result of that discussion was:

  • Without RTT estimation, the protocol runs in lock-step mode (NSTART=1).

comment:4 Changed 10 years ago by cabo@…

  • Resolution set to fixed
  • Status changed from new to closed

While we are still eagerly waiting for research that will underlie the definition of advanced congestion control schemes, section 4.7 of coap-12 is pretty solid as a basic congestion avoidance mechanism. It appears nothing more is needed in the base document. Therefore, I'm closing the ticket now.

Note: See TracTickets for help on using tickets.