发件人: Randriamasy, Sabine (Nokia - FR) [mailto:[email protected]] 发送时间: 2016年7月6日 21:49 收件人: Qin Wu; IETF ALTO 主题: RE: Metric redefintion issue in ALTO TE Metric draft
Hi all, To resume the discussions on ALTO TE metrics: applications want to have a sense of the network capabilities in terms of e.g. bandwidth, latency or hop count. ALTO could provide them in an e2e, space and time aggregated form and differs from what eg I2RS provides. [Qin]: Good point, I think the metrics we defined in this draft should not limited to TE metrics, we should extend it to any performance metrics we can get from various different data source. Another difference we can make is ALTO cost metrics can more focus on end to end path while I2RS not. One feature to specify carefully is the measurement time intervals: on one hand the server may indicate it to hint the accuracy of the data but if a metric comes from multiple sources with multiple measurement configurations it may be problematic. The only thing the ALTO Server may expose to the Client is the validity period of such a metric, either "constant" until the next update or calendared. The client will have to rely on that indication and trust the Server. [Qin]: Agree, I can add some text in the introduction to clarify this. Thanks! Are there any thoughts on this matter in the WG? Thanks, Sabine De : alto [mailto:[email protected]] De la part de EXT Qin Wu Envoyé : samedi 26 mars 2016 04:42 À : IETF ALTO Objet : [alto] Metric redefintion issue in ALTO TE Metric draft In IETF90 Toronto meeting, one question was raised on draft-wu-alto-te-metrics https://tools.ietf.org/html/draft-wu-alto-te-metrics-06 is whether we redefine metric defined somewhere else. Here is the clarification: I think this is not metric redefinition and what we do is to feed TE metric gathered using routing protocol into ALTO server and provide such TE metric to ALTO client. ALTO server may aggregate metric data gathered using various routing protocols. But the purpose is to normalize the data and provide it to ALTO client in more useful way, e.g., calculate end to end metric based on hop by hop metric and expose it to the ALTO client. -Qin
_______________________________________________ alto mailing list [email protected] https://www.ietf.org/mailman/listinfo/alto
