Dear AQMers, I write this email on the behalf of the co-authors of the draft "AQM evaluation guidelines". Based on the various feed-backs and discussions we could have had during and after the meeting, we would like to share to the lists our ideas for future directions.
One of the main issues of the draft that has been reported is that it tried to provide generic guidelines to evaluate an AQM AND guidelines to evaluate the AQM in specific context. Some discussions during the meeting were about the metrics for specific context, link bandwidth variations (WiFi), or ways to evaluate various metrics. These feed-backs and discussions illustrate that it may be too ambitious and confusing to propose both generic and context depend scenarios to evaluate AQM in one single document. In the current draft, there may be to much content making it difficult to come up with a consensus in the WG. Based on these discussions and feed-backs, we propose to split the current document into two drafts to ease the discussion: - *-aqm-eval-guidelines-*: this draft would provide high level generic guidelines to evaluate 'key' behavior of an AQM. This will include relevant discussions about metrics, topology and deployment - *-aqm-eval-suites-*: this draft would be an AQM evaluation suites which would consider specific networks conditions with respective specific metrics We believe that such splitting would let us integrate the various comments from the group and be more flexible for discussions and agreements on the content of each draft. We will try to submit an update as soon as possible and we welcome comments and discussions. Kind regards, The authors ----- Mail original ----- De: "Preethi Natarajan" <[email protected]> À: [email protected] Envoyé: Mercredi 5 Mars 2014 06:17:19 Objet: [aqm] Evaluation Guidelines: Next Steps Hello AQMers, Based on the discussions today at the WG session and other hallway conversations, it seems that the following are possible options: 1. Split the current version into multiple drafts as appropriate: The draft in its current version is bloated [aka "ContentBloat" :)], and can be logically separated out into multiple drafts. One way to split -- (I) a generic evaluations draft that focuses on a common denominator of goals/metrics expected out of an AQM scheme irrespective of deployment scenario (DC, 3G etc), and (ii) one or more drafts for specific network environments such as 3G/LTE, Data center discussing special goals and/or metrics. There could be other ways of splitting the draft, lets discuss those as well. Splitting gives a de-bloated generic evaluations draft that may be easier to process and get consensus on. WG can focus on special network scenarios drafts after that. 2. Generic evaluations draft contents/organization: Assuming (1), the next question is what should be contents of the generic draft. Based on discussions today, the draft should first discuss goals along with motivation for each goal, followed by metrics that make sense for each goal. I.e., metrics driven by the goals. The draft will also include generic evaluation criteria to evaluate an AQM against each goal. Please respond with comments/suggestions, Thanks, Preethi _______________________________________________ aqm mailing list [email protected] https://www.ietf.org/mailman/listinfo/aqm _______________________________________________ aqm mailing list [email protected] https://www.ietf.org/mailman/listinfo/aqm
