Greg – Inline.
From: Greg Mirsky <gregimir...@gmail.com> Sent: Monday, January 10, 2022 3:36 PM To: Les Ginsberg (ginsberg) <ginsb...@cisco.com> Cc: Tony Li <tony...@tony.li>; Christian Hopps <cho...@chopps.org>; Robert Raszuk <rob...@raszuk.net>; Aijun Wang <wangai...@tsinghua.org.cn>; Shraddha Hegde <shrad...@juniper.net>; Hannes Gredler <han...@gredler.at>; lsr <lsr@ietf.org>; Peter Psenak (ppsenak) <ppse...@cisco.com> Subject: Re: [Lsr] BGP vs PUA/PULSE Hi Les, thank you for the detailed clarifications. Please find my follow-up notes in-lined below under the GIM>> tag. Regards, Greg On Mon, Jan 10, 2022 at 3:19 PM Les Ginsberg (ginsberg) <ginsb...@cisco.com<mailto:ginsb...@cisco.com>> wrote: Greg – The obvious issue is scale. Since you need a full mesh you are talking about N**2 behavior – so it doesn’t take many nodes to require thousands of BFD sessions. GIM>> If I understand the scenario correctly, N represents the number of PEs, not the number of routers in ASes. If that is the case, what could be a good estimate for N? [LES:] Even a modest sized N = 100 (which is certainly not a high number) leads to 10000 BFD sessions. N= 500 => 250,000 sessions. Etc. In terms of detect time, we are trying to get an order of magnitude improvement from normal BGP session timers – so we are aiming for a modest number of seconds. GIM>> That is very helpful information, thank you. Then, we can expect that a one-second interval for the transmission of a BFD Control packet would be acceptable and guarantee failure detection within three seconds. If that is the case, I'll note that many platforms support thousands of BFD sessions at 3.3 msec intervals. It appears to me that the case we're discussing produces/processes 330 times fewer BFD packets per session. Should somewhat help with the scaling, would you agree? [LES:] Nodes which can support thousands of BFD sessions are likely already using many BFD sessions for other purposes. In particular, fast detection of local failures is always going to be a priority – so if a node has thousands of neighbors – it will likely have thousands of single hop BFD sessions. Not to mention the plethora of other OAM uses cases being defined. And the network-wide traffic impact as these new BFD sessions are largely multi-hop. Are you really arguing that the introduction of many thousands of BFD sessions is something we should not be concerned about? Les Les From: Greg Mirsky <gregimir...@gmail.com<mailto:gregimir...@gmail.com>> Sent: Monday, January 10, 2022 1:30 PM To: Les Ginsberg (ginsberg) <ginsb...@cisco.com<mailto:ginsb...@cisco.com>> Cc: Tony Li <tony...@tony.li<mailto:tony...@tony.li>>; Christian Hopps <cho...@chopps.org<mailto:cho...@chopps.org>>; Robert Raszuk <rob...@raszuk.net<mailto:rob...@raszuk.net>>; Aijun Wang <wangai...@tsinghua.org.cn<mailto:wangai...@tsinghua.org.cn>>; Shraddha Hegde <shrad...@juniper.net<mailto:shrad...@juniper.net>>; Hannes Gredler <han...@gredler.at<mailto:han...@gredler.at>>; lsr <lsr@ietf.org<mailto:lsr@ietf.org>>; Peter Psenak (ppsenak) <ppse...@cisco.com<mailto:ppse...@cisco.com>> Subject: Re: [Lsr] BGP vs PUA/PULSE Hi Les, thank you for bringing the real-life scenarios to the discussion. In your opinion, what prevents an operator from monitoring a remote PE using a multi-hop BFD? Do you have an estimated number of such sessions each PE must handle? What could be the required guaranteed failure detection time? Best regards, Greg On Mon, Jan 10, 2022 at 1:08 PM Les Ginsberg (ginsberg) <ginsberg=40cisco....@dmarc.ietf.org<mailto:40cisco....@dmarc.ietf.org>> wrote: Chris/Tony – We have received requests from real customers who both need to summarize AND would like better response time to loss of reachability to individual nodes. If they could operate at the necessary scale without summarizing they would have already – so telling customers to simply make sure they don’t use summaries isn’t helpful. There are then two ways to respond: 1)Sorry, when you use summaries you lose the ability to receive state information about individual prefixes covered by the summary. There is nothing we can do to help you. This seems to be what the two of you are saying. 2)We can provide a way to improve response time for the loss of reachability to individual destinations covered by a summary, but its use will be limited to isolated failures. Failures which affect a significant number of destinations at the same time will realize no benefit from the solution. If this limitation is acceptable then we have proposals that we think will be useful. That’s what we are trying to do. Les From: Tony Li <tony1ath...@gmail.com<mailto:tony1ath...@gmail.com>> On Behalf Of Tony Li Sent: Monday, January 3, 2022 1:09 PM To: Christian Hopps <cho...@chopps.org<mailto:cho...@chopps.org>> Cc: Peter Psenak (ppsenak) <ppse...@cisco.com<mailto:ppse...@cisco.com>>; Les Ginsberg (ginsberg) <ginsb...@cisco.com<mailto:ginsb...@cisco.com>>; Robert Raszuk <rob...@raszuk.net<mailto:rob...@raszuk.net>>; Shraddha Hegde <shrad...@juniper.net<mailto:shrad...@juniper.net>>; Aijun Wang <wangai...@tsinghua.org.cn<mailto:wangai...@tsinghua.org.cn>>; Hannes Gredler <han...@gredler.at<mailto:han...@gredler.at>>; lsr <lsr@ietf.org<mailto:lsr@ietf.org>> Subject: Re: [Lsr] BGP vs PUA/PULSE On Jan 3, 2022, at 11:23 AM, Christian Hopps <cho...@chopps.org<mailto:cho...@chopps.org>> wrote: And I'm saying if a prefix is important enough to merit a bunch of new protocol extensions and state, then it's important enough to simply be left out of the summarization in the first place. And then people get what they want, w/o protocol changes/upgrades, and it's using time tested and hardened IGP code and designs. +1 T _______________________________________________ Lsr mailing list Lsr@ietf.org<mailto:Lsr@ietf.org> https://www.ietf.org/mailman/listinfo/lsr
_______________________________________________ Lsr mailing list Lsr@ietf.org https://www.ietf.org/mailman/listinfo/lsr