Last Call Review of draft-ietf-tsvwg-circuit-breaker-11
review-ietf-tsvwg-circuit-breaker-11-opsdir-lc-dunbar-2016-02-13-00
Request | Review of | draft-ietf-tsvwg-circuit-breaker |
---|---|---|
Requested revision | No specific revision (document currently at 15) | |
Type | Last Call Review | |
Team | Ops Directorate (opsdir) | |
Deadline | 2016-02-09 | |
Requested | 2016-02-03 | |
Authors | Gorry Fairhurst | |
I-D last updated | 2016-02-13 | |
Completed reviews |
Genart Last Call review of -11
by Roni Even
(diff)
Genart Telechat review of -13 by Roni Even (diff) Secdir Last Call review of -11 by Benjamin Kaduk (diff) Opsdir Last Call review of -11 by Linda Dunbar (diff) Opsdir Telechat review of -13 by Linda Dunbar (diff) Rtgdir Early review of -11 by Andrew G. Malis (diff) |
|
Assignment | Reviewer | Linda Dunbar |
State | Completed | |
Request | Last Call review on draft-ietf-tsvwg-circuit-breaker by Ops Directorate Assigned | |
Reviewed revision | 11 (document currently at 15) | |
Result | Serious Issues | |
Completed | 2016-02-13 |
review-ietf-tsvwg-circuit-breaker-11-opsdir-lc-dunbar-2016-02-13-00
I have reviewed this document as part of the Operational directorate's ongoing effort to review all IETF documents being processed by the IESG. These comments were written with the intent of improving the operational aspects of the IETF drafts. Comments that are not addressed in last call may be included in AD reviews during the IESG review. Document editors and WG chairs should treat these comments just like any other last call comments. Major issue with the proposed approach: Traffic continuously go from Ingress and egress. If the Ingress and egress nodes don’t have synchronized timers as SONET/SDH (most packet routers/switches don’t have synchronized timer), the meters collected by both ends can’t be correlated properly (i.e. they are meaningless). The only ways for the meters on both Ingress and Egress to work reliably (without synchronized timer) is via ECN, or via synthetic data (like IPPM). For “Circuit Breaker based on RTP”, DPI has to be deployed to detect RTP if the Ingress and egress nodes are the end points of tunnel (i.e. aggregated traffic). The approach is very expensive. Minor issue: Typos: - Page 5: The paragraph before 1.1: “Circuit Breaker has in fact being tripped”: do you mean “being triggered”? Regards, Linda Dunbar