Dear Greg,

(I re-shaped the title for more explicit indication of the topic.)
Thanks for your comments, and sorry for my delayed response.
Please see my replies inline.

From: Greg Mirsky [mailto:[email protected]]
Sent: Monday, July 22, 2019 11:52 PM
To: [email protected]; RTGWG <[email protected]>
Subject: Followup on my comments

Dear Authors, et al.,
to summarize my comments at the mike:

  *   BFD WG will discuss possible extensions to BFD proposed in 
draft-mirmin-bfd-extended<https://tools.ietf.org/html/draft-mirmin-bfd-extended-01>.
 Appreciate your comments to the potential use of the proposed ideas for iCAN.
[Bing] Please let me clarify that iCAN was not designed to compete with BFD. 
They are basically two different things. iCAN is regarding to allowing the 
routers switch flows between multiple paths autonomously without the 
involvement of the Controller. In order to do this, the routers also need to do 
some measurement of the paths. Thus, for the scenarios that already deploy 
iCAN, we thought it won’t be necessary to enable BFD for the path detection. 
So, this is a “bonus” for the iCAN scenarios.
For general scenarios, especially link failure detection, I believe BFD is 
still inevitable.

  *   It is my understanding that iCAN enables transient nodes to measure 
Residence time as defined in Section 2 RFC 
8169<https://tools.ietf.org/html/rfc8169#section-2>. Is my understanding of 
iCAN correct?
  *   in section 4.4 of 
draft-mirsky-sfc-pmamm<https://tools.ietf.org/html/draft-mirsky-sfc-pmamm-08> 
we discuss how the Alternate Marking method can be used to measure the 
Residence Time for an SFF and/or an SF. That can be extended to other overlay 
and underlay networks.
[Bing] iCAN is not aimed for measuring Residence Time; it is about “how 
congested the path is?”, which is basically a ratio of the 
Egress-Receiving-Throughput and Ingress-Sending-Throughput on a specific path. 
In this way, we can get very instant measurement result for very fast flow path 
adjustment (in our current prototype, it is done every 10ms.).
Best regards,
Bing

Regards,
Greg
_______________________________________________
rtgwg mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/rtgwg

Reply via email to