Last Call Review of draft-ietf-tcpm-accecn-reqs-07
review-ietf-tcpm-accecn-reqs-07-opsdir-lc-dunbar-2014-12-04-00
| Request | Review of | draft-ietf-tcpm-accecn-reqs |
|---|---|---|
| Requested revision | No specific revision (document currently at 08) | |
| Type | Last Call Review | |
| Team | Ops Directorate (opsdir) | |
| Deadline | 2014-12-04 | |
| Requested | 2014-11-21 | |
| Authors | Mirja Kühlewind , Richard Scheffenegger , Bob Briscoe | |
| I-D last updated | 2014-12-04 | |
| Completed reviews |
Genart Last Call review of -07
by Brian E. Carpenter
(diff)
Genart Telechat review of -07 by Brian E. Carpenter (diff) Opsdir Last Call review of -07 by Linda Dunbar (diff) |
|
| Assignment | Reviewer | Linda Dunbar |
| State | Completed | |
| Request | Last Call review on draft-ietf-tcpm-accecn-reqs by Ops Directorate Assigned | |
| Reviewed revision | 07 (document currently at 08) | |
| Result | Has issues | |
| Completed | 2014-12-04 |
review-ietf-tcpm-accecn-reqs-07-opsdir-lc-dunbar-2014-12-04-00
As a member of the Operations Directorate, I was asked to review draft-ietf-tcpm-accecn-reqs-07. In general, the draft is written very well. But I think the “requirements” listed in Section 4 is more like “design consideration”. For example, the proposed solution in 5.1 doesn’t meet the “resilience” requirement listed in the Section 4. On the Page 8 under “Timeliness”, the text says “… should be delivered within about one RTT”. But the earlier sentence describes that it takes longer than one RTT for the ACK to reach the sender. Here it says that "should be within one RTT". Is there any particular value to set for a specific attribute to reach this goal? On the Page 9 under “Accuracy”, the text says: “In the best case the new scheme should even allow reconstruction of the exact number of payload bytes that a CE marked packet was carrying”. Why not using the value in the Length field of the IP header? Cheers, Linda Dunbar