Hi Haoyu,
I am interested in your thoughts on the two closed loops. I agree with you. I think it's useful to clarify this point in the draft. The per packet hop by hop monitoring will cost too much resource. So the solution we are looking for is iterative. Spatially, we can monitor selected services or tunnels at one time, and switch to other monitoring objects then. Temporally, we can do the end to end or sampled monitoring, and switch to the fine grained monitoring once the SLA degrade. And so the telemetry could be interactive. For example, when we detect the end to end packet loss, we firstly evaluate the frequency, the pattern and see if it's caused by some obvious reason (e.g., the power failure). Then we decide whether to apply the fine grained monitoring and locate the fault. I think these could all be implemented in a closed loop telemetry. Best Regards, Fengwei Qin 发件人: Haoyu Song [mailto:[email protected]] 发送时间: 2019年9月21日 01:30 收件人: qinfengwei; [email protected]; [email protected] 主题: RE: [OPSAWG] iFIT framework draft Hi Fengwei, Thanks for the feedback. My understanding is that there are two types of closed loops. In our draft, iFIT’s closed loop is for changing the telemetry requirements based on previous telemetry results, so the whole loop is still for the telemetry purpose. Another type of closed loop is so called closed control loop, which means, based on the telemetry results, the application will automatically change the network policy or configuration. For this type, iFIT is just part of the loop and we may not want to cover that because it might belong to a bigger topic (such as automatic network) . But I think it’s useful to clarify this point in the draft. The “self-healing” case you mentioned might belong to the second type, but the “precision location of silent failures” case might belong to the first type, depending on a better understanding of the case, i.e., what kind of failures and how data-path telemetry can help to detect. If you and anybody else have such information, I would like to hear and include the discussion in the draft. Thanks! Haoyu From: qinfengwei <[email protected]> Sent: Friday, September 20, 2019 7:09 AM To: Haoyu Song <[email protected]>; [email protected]; [email protected] Subject: 答复: [OPSAWG] iFIT framework draft Hi Haoyu, I have read the draft and think the iFIT is useful for operators to identify the issue quickly. You know that the silent failures are our biggest pain points, the precision location and network self-healing are our rigid demand. You say that iFIT is a closed-loop framework, in my opinion, it should include how to diagnose and restore the network, but I haven't seen either. Do I not understand or would you have explained it in other draft? Beat Regard, Fengwei Qin 发件人: OPSAWG [mailto:[email protected]] 代表 Haoyu Song 发送时间: 2019年7月18日 02:06 收件人: [email protected]; [email protected] 主题: [OPSAWG] iFIT framework draft Hi all, We recently uploaded a newer version of the iFIT framework draft. https://datatracker.ietf.org/doc/draft-song-opsawg-ifit-framework/ <https://nam03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdatatracker.ietf.org%2Fdoc%2Fdraft-song-opsawg-ifit-framework%2F&data=02%7C01%7Chaoyu.song%40futurewei.com%7C22e31da5670b48dcbbea08d73dd43232%7C0fee8ff2a3b240189c753a1d5591fedc%7C1%7C0%7C637045853915262328&sdata=7dKlBcMx%2FyrDVSPNUOmG%2BaC0x4AfRGVQvgoAWmmyEYc%3D&reserved=0> The purposes of the drafts are: 1. Clarify the terms and underlying techniques for data plane on-path telemetry 2. Present a framework that addresses the practical implementation and deployment issues 3. Identify the open issues and directions for related standard development Given the wide industry interests on adopting such techniques, we believe this informational draft is useful and In the scope of OPSAWG. Please review the draft and provide your opinions, suggestions, and comments. Thank you very much! Haoyu
_______________________________________________ OPSAWG mailing list [email protected] https://www.ietf.org/mailman/listinfo/opsawg
