Speaking WG member: Hi Gyan, Aijun,
Even if I agreed with your use case assuming a passive interface denotes the boundary between two domains as shown in figure 1 in your draft (which I don’t), you still should not be trying to hack the prefixes with what is inherently link attribute. Can I state this anymore simply or are you are just going to restate your requirements again??? Acee From: Gyan Mishra <[email protected]> Date: Sunday, October 11, 2020 at 3:19 PM To: Acee Lindem <[email protected]> Cc: Aijun Wang <[email protected]>, Aijun Wang <[email protected]>, "Peter Psenak (ppsenak)" <[email protected]>, "[email protected]" <[email protected]> Subject: Re: [Lsr] FW: New Version Notification for draft-wang-lsr-passive-interface-attribute-04.txt Hi Acee I believe what Aijun is trying to explain is that the primary purpose of PCE for active or passive path computation is for inter-as RSVP-TE PCALC path computation or SR-TE path computation. So PCE is solving a well known PCALC bin packing problem due to pcalc over subscribing RSVP tunnel bandwidth which is inherent an RSVP issue, but a bigger problematic issue is with being able to build an inter-as TE path with a single or multiple PCE controllers that can take the LSDB link attributes in the TEDs dB opaque LSAs in the ospf case and ISIS TE TLVs and rebuild the topology topology from each TE domain to be able to build a congruent end to end RSVP TE or SR-TE traffic steered path instantiation. Without the use of PCE controllers as the LSDB link attribute information is not known as RSVP loose ERO static lsp is built or SR SR-TE prefix sid loose path is built, failover due to crank back is impacted for reroute, due to not having a complete depiction of the other AS Link state topology by the head end or SR source node. So to build that entire end to end inter-as path for that end to end RSVP TE or SR-TE path instantiation it is critical to indentify the inter-as link eBGP tie link that may have static routes or BGP LU for RSVP head end to tail end reachability and SR-TE reachability to build out the end to end path instantiation. So the BGP-LS task to rebuild the lsdb topology of each inter connected AS that the RSVP TE or SR-TE steered path traverses, we need the accurate depiction and that includes the Identification of the critical inter-as tie link eBGP peering link that is passive for the PCE path computation logic for the end to end inter-as path instantiation. As for other interfaces using passive in the context of a operator service provider or enterprise core P and PE routers all links are transit with neighbors except the inter-as tie so having this new IGP TLV will help to that end. In the operator “core” network if there are other interfaces that don’t need to be advertised as stub, they can easily be excluded from being added into IGP. Kind Regards Gyan On Sat, Oct 10, 2020 at 6:29 AM Acee Lindem (acee) <[email protected]<mailto:[email protected]>> wrote: Hi Aijun, From: Aijun Wang <[email protected]<mailto:[email protected]>> Date: Friday, October 9, 2020 at 9:58 PM To: Acee Lindem <[email protected]<mailto:[email protected]>>, "Peter Psenak (ppsenak)" <[email protected]<mailto:[email protected]>>, 'Aijun Wang' <[email protected]<mailto:[email protected]>> Cc: "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>> Subject: RE: [Lsr] FW: New Version Notification for draft-wang-lsr-passive-interface-attribute-04.txt Hi, Acee: From: [email protected]<mailto:[email protected]> [mailto:[email protected]<mailto:[email protected]>] On Behalf Of Acee Lindem (acee) Sent: Saturday, October 10, 2020 3:48 AM To: Aijun Wang <[email protected]<mailto:[email protected]>>; Peter Psenak (ppsenak) <[email protected]<mailto:[email protected]>>; 'Aijun Wang' <[email protected]<mailto:[email protected]>> Cc: [email protected]<mailto:[email protected]> Subject: Re: [Lsr] FW: New Version Notification for draft-wang-lsr-passive-interface-attribute-04.txt Speaking as WG member: Hi Aijun, From: Aijun Wang <[email protected]<mailto:[email protected]>> Date: Thursday, October 8, 2020 at 11:09 PM To: Acee Lindem <[email protected]<mailto:[email protected]>>, "Peter Psenak (ppsenak)" <[email protected]<mailto:[email protected]>>, 'Aijun Wang' <[email protected]<mailto:[email protected]>> Cc: "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>> Subject: RE: [Lsr] FW: New Version Notification for draft-wang-lsr-passive-interface-attribute-04.txt Hi, Acee: Sorry for the previous pruned mail. Let's reply you again along your original question. Please see inline.[WAJ] -----Original Message----- From: [email protected]<mailto:[email protected]> [mailto:[email protected]] On Behalf Of Acee Lindem (acee) Sent: Wednesday, September 30, 2020 7:47 PM To: Aijun Wang <[email protected]<mailto:[email protected]>>; Peter Psenak (ppsenak) <[email protected]<mailto:[email protected]>>; 'Aijun Wang' <[email protected]<mailto:[email protected]>> Cc: [email protected]<mailto:[email protected]> Subject: Re: [Lsr] FW: New Version Notification for draft-wang-lsr-passive-interface-attribute-04.txt Hi Aijun, Other than your ill-conceived topology discovery heuristic [WAJ] The topology discovery heuristic is not suitable for the corner use case when the unnumbered interface is used, as explained in https://tools.ietf.org/html/draft-ietf-lsr-ospf-prefix-originator-06#appendix-B. If you don’t agree, would you like to illustrate other non-applicable scenarios? Right – and nobody other than yourself believes the IGPs should be modified to expose the abstracted topology of an area outside that area. [WAJ] The modification doesn’t change the way and complexity of route calculation within IGP. It just piggyback some extra information, the bulk of the reconstruction work is done by the controller. Such extra information can also have other usage, as described in https://tools.ietf.org/html/draft-ietf-lsr-ospf-prefix-originator-06#section-1 And, the proposal described in https://datatracker.ietf.org/doc/html/draft-wang-lsr-passive-interface-attribute-04 is different with the problem you concerned. It has no relation with the abstracted topology of an area. Maybe you are confused by these two drafts? It is a similar problem. You are still trying to overload the prefix advertisements with a link attribute (passive interface) so that it can be conveyed outside the domain. We certainly wouldn’t waste a limited prefix flag on this parochial application. You can solve the problem with BGP-LS session(s) between the router with a BGP-LS session to the controller and a router in each area w/o a router with a BGP-LS session to the controller. [WAJ] This is possible, but not efficient. For operation, we must also consider the configuration/administration overhead. BGP-LS is designed mainly for the northbound protocol, not east-west protocol. what other possible reason would there be for associating the passive attribute with a prefix? [WAJ] To know the boundary of the IGP domain. After knowing the boundary, the controller can safely apply and check the network security policy, the inbound traffic control policy etc. It really isn’t relevant, but I have to ask…. How does the presence of a prefix associated with a passive interface allow you to make this deduction? [WAJ] Passive interfaces are deployed mainly at the boundary of IGP domain. Is there any other exception? While passive interfaces are not standardized, there is nothing that limits their usage to an IGP boundary. They can and are deployed on any interface where adjacencies are not to be formed (e.g., a stub subnet containing hosts). Thanks, Acee Thanks, Acee Thanks, Acee On 9/29/20, 10:39 PM, "Aijun Wang" <[email protected]<mailto:[email protected]>> wrote: Hi, Acee and Peter: Passive interface is mainly used at the edge of the network, where the unnumbered interface will not be used. And the information to flag the passive interfaces is for positioning the area boundary, not conflict with the abstract capabilities of the area inside. Best Regards Aijun Wang China Telecom -----Original Message----- From: [email protected]<mailto:[email protected]> [mailto:[email protected]] On Behalf Of Acee Lindem (acee) Sent: Tuesday, September 29, 2020 9:16 PM To: Peter Psenak <[email protected]<mailto:[email protected]>>; Aijun Wang <[email protected]<mailto:[email protected]>>; 'Aijun Wang' <[email protected]<mailto:[email protected]>> Cc: [email protected]<mailto:[email protected]> Subject: Re: [Lsr] FW: New Version Notification for draft-wang-lsr-passive-interface-attribute-04.txt Speaking as WG member: Hi Aijun, Peter, I agree with Peter - one of the main motivations for having areas is to abstract the topology within the area. Now you're trying to supplant this - one topological detail at a time with ill-conceived IGP features. Thanks, Acee On 9/29/20, 5:15 AM, "Lsr on behalf of Peter Psenak" <[email protected] on behalf of [email protected]<mailto:[email protected]%20on%20behalf%20of%[email protected]>> wrote: Hi Aijun, On 29/09/2020 11:07, Aijun Wang wrote: > Hi, Peter: > > Thanks for your comments. > 1. For BGP-LS deployment, there normally only be one router that within the > IGP domain to report the topology information, this router should know such > passive links which exists mainly on other border routers via the IGP > protocol. This is main reason to extension the IGP protocol. > 2. For the solution, normally, the link within the IGP connect two ends, but > passive interface is special and not fall in this space. We have studied the > current TLVs that for link, and find no suitable container to append this > information. This is the reason that we select the TLVs that associated with > Prefix. if the link is unnumbered, your solution does not work. As I said, if you need a knowledge about the link, you can not advertise it as a prefix. thanks, Peter > >>From other POV, the OSPFv3 defines now the "Intra-Area-Prefix LSA", which > isolate the prefix information that associated with link into this > container, contains the stub link, local interface information etc. Put such > attribute along with the prefix is then acceptable? > > > Best Regards > > Aijun Wang > China Telecom > > -----Original Message----- > From: [email protected]<mailto:[email protected]> [mailto:[email protected]] On Behalf Of Peter > Psenak > Sent: Tuesday, September 29, 2020 4:29 PM > To: Aijun Wang <[email protected]<mailto:[email protected]>> > Cc: [email protected]<mailto:[email protected]> > Subject: Re: [Lsr] FW: New Version Notification for > draft-wang-lsr-passive-interface-attribute-04.txt > > Hi Aijun, > > here's my comments: > > The purpose of this draft is to advertise passive links. > > 1. I'm not sure the problem needs to be solved by IGPs. I tend to believe > ietf-idr-bgpls-inter-as-topology-ext is sufficient. > > 2. the solution that you proposed is wrong. You are trying to derive > topological data about the passive links from the prefix advertisement. > This is semantically incorrect and only works under very specific condition. > If you need to advertise a link, advertise it as a "special" > link, not as a "special" prefix. > > thanks, > Peter > > On 29/09/2020 03:17, Aijun Wang wrote: >> Hi, Peter: >> >> Would you like to review and give comments on the updates version of this > draft? >> We have also added the protocol extension proposal for OSPFv3. >> >> The update version of this draft can refer to >> https://datatracker.ietf.org/doc/html/draft-wang-lsr-passive-interface >> -attribute >> Thanks in advance. >> >> >> Best Regards >> >> Aijun Wang >> China Telecom >> >>> -----Original Message----- >>> From: [email protected]<mailto:[email protected]> [mailto:[email protected]] >>> Sent: Monday, September 28, 2020 3:17 PM >>> To: Zhibo Hu <[email protected]<mailto:[email protected]>>; Gyan Mishra >>> <[email protected]<mailto:[email protected]>>; Aijun Wang <[email protected]<mailto:[email protected]>>; >>> Gyan S. Mishra <[email protected]<mailto:[email protected]>> >>> Subject: New Version Notification for >>> draft-wang-lsr-passive-interface-attribute-04.txt >>> >>> >>> A new version of I-D, >>> draft-wang-lsr-passive-interface-attribute-04.txt >>> has been successfully submitted by Aijun Wang and posted to the IETF >>> repository. >>> >>> Name: draft-wang-lsr-passive-interface-attribute >>> Revision: 04 >>> Title: Passive Interface Attribute >>> Document date: 2020-09-28 >>> Group: Individual Submission >>> Pages: 7 >>> URL: >>> https://www.ietf.org/id/draft-wang-lsr-passive-interface-attribute-04. >>> txt >>> Status: >>> https://datatracker.ietf.org/doc/draft-wang-lsr-passive-interface-att >>> r >>> ibute/ >>> Htmlized: >>> https://datatracker.ietf.org/doc/html/draft-wang-lsr-passive-interfac >>> e >>> -attribut >>> e >>> Htmlized: >>> https://tools.ietf.org/html/draft-wang-lsr-passive-interface-attribut >>> e >>> -04 >>> Diff: >>> https://www.ietf.org/rfcdiff?url2=draft-wang-lsr-passive-interface-at >>> t >>> ribute-04 >>> >>> Abstract: >>> This document describes the mechanism that can be used to >>> differentiate the passive interfaces from the normal interfaces >>> within ISIS or OSPF domain. >>> >>> >>> >>> >>> Please note that it may take a couple of minutes from the time of >>> submission until the htmlized version and diff are available at > tools.ietf.org<http://tools.ietf.org>. >>> >>> The IETF Secretariat >>> >> >> >> >> > > _______________________________________________ > Lsr mailing list > [email protected]<mailto:[email protected]> > https://www.ietf.org/mailman/listinfo/lsr > > > _______________________________________________ Lsr mailing list [email protected]<mailto:[email protected]> https://www.ietf.org/mailman/listinfo/lsr _______________________________________________ Lsr mailing list [email protected]<mailto:[email protected]> https://www.ietf.org/mailman/listinfo/lsr _______________________________________________ Lsr mailing list [email protected]<mailto:[email protected]> https://www.ietf.org/mailman/listinfo/lsr _______________________________________________ Lsr mailing list [email protected]<mailto:[email protected]> https://www.ietf.org/mailman/listinfo/lsr -- [Image removed by sender.]<http://www.verizon.com/> Gyan Mishra Network Solutions Architect M 301 502-1347 13101 Columbia Pike Silver Spring, MD
_______________________________________________ Lsr mailing list [email protected] https://www.ietf.org/mailman/listinfo/lsr
