Terminology for Benchmarking Link-State IGP Data-Plane Route Convergence
Note: This ballot was opened for revision 23 and is now closed.
(David Ward) Discuss
Discuss (2007-07-17 for -** No value found for 'p.get_dochistory.rev' **)
A few items: 0) we need to have a definition of remote vs local failure 1) Why isn't a convergence event defined as any local or remote trigger that causes a route recalculation vs one in which fwding is effected. If convergence is only to be defined by a change in forwarding what is the term that the authors recommend for an event in which a route calculation has to be made but, in fact forwarding is not changed? To the control plane of the router, the work is the same and given a catastrophic network event; a "queue" of calculations that cause no forwarding change in front of a calculation that would cause a forwarding plane change is critical to define, understand and place as a variable in the convergence equation. 2) There should be a definition of prioritized convergence in which "important prefixes" (e.g. loopbacks that are BGP NHs) are measured vs "unimportant prefixes." In addition, the important prefixes should 3) There have been alternative definitions and terminology for convergence that the authors should cite and rectify. Many of these docs have been discussed in rtgwg. 4) loops and microloops should be defined 5) units of measurement are wrong order of magnitude 6) Restoration Convergence time is unclear. The IGP sees only individual convergence events.
(Ron Bonica) Yes
(Jari Arkko) No Objection
(Stewart Bryant) (was Discuss) No Objection
(Ross Callon) No Objection
(Gonzalo Camarillo) No Objection
(Ralph Droms) No Objection
Comment (2010-07-01 for -** No value found for 'p.get_dochistory.rev' **)
In several definitions: Measurement Units: hh:mm:ss:nnn:uuu, where 'nnn' is milliseconds and 'uuu' is microseconds. The "Measurement Units" are microseconds, while "hh:mm:ss:nnn:uuu" is a representation. Elsewhere, "Measurement Units" are defined as, e.g., "seconds" I don't understand the requirements language (this example from section 3.1.2): Discussion: Full Convergence MUST occur after a Convergence Event. "MUST occur" for compliance or interoperability with what, exactly?
(Lisa Dusseault) No Objection
(Lars Eggert) No Objection
(Adrian Farrel) No Objection
Comment (2010-06-30 for -** No value found for 'p.get_dochistory.rev' **)
Support Stewart Bryant's Discuss
(Stephen Farrell) No Objection
(Sam Hartman) No Objection
(Russ Housley) No Objection
Comment (2007-07-14 for -** No value found for 'p.get_dochistory.rev' **)
Based on Gen-ART Review by Vijay K. Gurbani. This draft is ready for publication as an Informational RFC. There are a few editorial changes that should be made: - Section 1: s/of the DUT and the/of the DUT, and the/ - Section 3.1: DUT is expanded here; if it should be expanded anywhere, it should be in Section 1.
(Cullen Jennings) No Objection
(Chris Newman) No Objection
(Tim Polk) No Objection
(Pete Resnick) No Objection
Like Ralph, I am very confused by the use of 2119 language in this document. I don't understand what its necessity is or who it is aimed at.
(Dan Romascanu) No Objection
Comment (2007-07-18 for -** No value found for 'p.get_dochistory.rev' **)
In section 3.5 I do not understand why the measurment unit reads: 'Number of N-octet offered packets that are not forwarded' Why not just? 'Number of packets that are not forwarded' If the definition is packet loss for a packet of length N, then it is the definition field that needs to be changed.