Re: [Isis-wg] Conflicting MS entries

Hi Uma, 

Pls find inline comments.

-----Original Message-----
From: Uma Chunduri [mailto:uma.chunduri@ericsson.com] 
Sent: Tuesday, June 30, 2015 20:24
To: Aissaoui, Mustapha (Mustapha); DECRAENE Bruno IMT/OLN
Cc: LITKOWSKI Stephane SCE/IBNF; spring@ietf.org; isis-wg@ietf.org list; Stefano Previdi (sprevidi)
Subject: RE: Conflicting MS entries

Installing multiple SIDs for the same prefix in the FWD is bit off topic though related to the point being discussed.

Have few questions here to answer the "Conflicting MS Entries"

1. Is it a valid configuration to  provision multiple global prefix SIDs for any prefix  ?
     It seems the consensus here is though one can do this there is no real use case.
[SLI] For local prefixes, I agree that there would be no use case for this (for now).
But two different nodes can advertise the same SID for different prefix, or advertise mapping entries for the same prefix (using different range semantics) with different SIDs and it s harder to detect.

2. Is it a valid configuration to host a prefix on two nodes with different global prefix SIDs?
     In the similar lines is it valid a configuration to host multiple MSes and advertise different global SID indices for the same prefix?
    I failed to see any valid use case here too.
[SLI] Agree, but even if there is no use case, such configuration can be done by human error, so we need to have a consistent behavior.

3. Is it a valid configuration to host a connected or otherwise route on two nodes with same prefix SID?

To me answer for #1 and #2 is invalid.  
 #3 is valid and intention is to host a multi-homed prefix so that all nodes can  compute ECMP path or prefer lower cost path to one of the nodes.
[SLI] Agree, the inter-area case will also provide this kind of scenario.

I prefer staying out of defining any mechanism  which would be any ways  indeterministic for invalid configurations. 

--
Uma C.

-----Original Message-----
From: spring [mailto:spring-bounces@ietf.org] On Behalf Of Aissaoui, Mustapha (Mustapha)
Sent: Wednesday, June 24, 2015 6:37 AM
To: bruno.decraene@orange.com
Cc: LITKOWSKI Stephane SCE/IBNF; spring@ietf.org; isis-wg@ietf.org list; Stefano Previdi (sprevidi)
Subject: Re: [spring] Conflicting MS entries

Hi Bruno,
Indeed what you are describing would be similar to having tunnels to the same destination prefix resolved via two different protocols (LDP and SR). As you said, maybe I am just not seeing a use case for allowing this. I can however see that one can program two SIDs for the same prefix using two different SPF algorithms.

Regards,
Mustapha.

> -----Original Message-----
> From: spring [mailto:spring-bounces@ietf.org] On Behalf Of 
> bruno.decraene@orange.com
> Sent: Wednesday, June 24, 2015 3:38 AM
> To: Aissaoui, Mustapha (Mustapha)
> Cc: LITKOWSKI Stephane SCE/IBNF; spring@ietf.org; isis-wg@ietf.org 
> list; Stefano Previdi (sprevidi)
> Subject: Re: [spring] Conflicting MS entries
> 
> Hi Mustapha,
> 
> > From: Aissaoui, Mustapha (Mustapha)
> > [mailto:mustapha.aissaoui@alcatel-
> > lucent.com]
> > Sent: Tuesday, June 23, 2015 10:34 PM
> >
> > Hi Bruno,
> > I can see that one can program multiple SIDs if these are for routes 
> > of the same prefix which are advertised in multiple ISIS instances 
> > or in multiple topologies. But within a single instance and 
> > topology, how do you reconcile the multiple MS entries with a single 
> > active route to that
> destination prefix?
> 
> I'm probably missing something, but I don't see a need to reconcile. I 
> see 2 LSPs and for each a (N:1) indirection between the (label, FEC
> element) and the IP FIB. A bit similar to multiple BGP routes using 
> the same IP prefix to resolve their BGP Next-Hop.
> 
> e.g. MS advertises 2 SIDs for prefix1 --> (via SRGBs) 2 incoming & 2 
> outgoings labels. LL1, LL2 for local/incoming labels. LN1, LN2 for neighbor/outgoing label.
> IP FIB:
> Prefix1 --> eth1
> 
> NHLFE:
> LFE1: eth1, swap LN1
> LFE2: eth1, swap LN2
> 
> ILM:
> LL1 --> LFE1
> LL2 --> LFE2
> 
> 
> Looks also similar to the case where a node(LSR) is both SR & LDP and 
> for the same FEC element/IP prefix  gets 1 label from LDP and 1 from SR/MS.
> 
> That being said, if we are all in favor of selecting a single MS 
> entry, the discussion is purely theoretical.
> 
> /Bruno
> 
> > Mustapha.
> >
> > > -----Original Message-----
> > > From: spring [mailto:spring-bounces@ietf.org] On Behalf Of 
> > > bruno.decraene@orange.com
> > > Sent: Tuesday, June 23, 2015 9:48 AM
> > > To: Aissaoui, Mustapha (Mustapha)
> > > Cc: LITKOWSKI Stephane SCE/IBNF; spring@ietf.org; isis-wg@ietf.org 
> > > list; Stefano Previdi (sprevidi)
> > > Subject: Re: [spring] Conflicting MS entries
> > >
> > > Hi Mustapha,
> > >
> > > Thanks for the discussion. Please see inline.
> > >
> > > > From: Aissaoui, Mustapha (Mustapha) 
> > > > [mailto:mustapha.aissaoui@alcatel-> lucent.com] > Sent: Friday, 
> > > > June 19, 2015 8:19 PM
> > > >
> > > > Hi Stephane and Bruno,
> > > > I do not think programming multiple SIDs makes sense. While 
> > > > there are multiple MS prefix sub-TLVs, there is only single 
> > > > active route for the prefix with potentially ECMP next-hops 
> > > > which was resolved from a received IP reachability TLV.
> > >
> > > I don't see what prevents us from programming multiple SIDs/labels 
> > > for a single prefix. i.e. setting up multiple LSPs for a given prefix.
> > > e.g. BGP seems to specifically allow for this
> > > https://tools.ietf.org/html/rfc3107#section-4
> > >
> > > That being said:
> > > - I'm not seeing benefit (which, may be, is what you meant by 
> > > "makes no sense")
> > > - Stéphane expressed that from an operational point, this may make 
> > > things
> > harder.
> > > - eventually some implementation may find this harder compared to 
> > > having a single one.
> > >
> > > So, so far, it looks like everyone expressed a preference to use 
> > > only one, based on deterministic criteria.
> > >
> > > > I agree that selecting one of the entries is preferable to dropping traffic.
> > > Ok.
> > >
> > > > We
> > > > could come up with selection criteria but the reality is that 
> > > > there no way for the router to check if any of the MS entries is 
> > > > legitimate or
> not.
> > >
> > > What do you mean by "legitimate"?
> > > In case of multiple SID advertisement for a prefix, they seem all 
> > > equally legitimate to me. Looks like giving multiple names to 
> > > something. e.g. we'll call this LSP/segment "A" or "R" or "Y".
> > > Now we may choose to only use one, based on a criteria TBD. e.g.
> > > smallest SID (which a priori improve the probability of fitting 
> > > inside the
> > SRGBs).
> > >
> > > /Bruno
> > >
> > > > As a result, I
> > > > would think that once an entry is selected based on the criteria 
> > > > and programmed, we should not be changing it unless the MS entry 
> > > > is
> > withdrawn.
> > > >
> > > > Mustapha.
> > > >
> > > > > -----Original Message-----
> > > > > From: spring [mailto:spring-bounces@ietf.org] On Behalf Of 
> > > > > stephane.litkowski@orange.com
> > > > > Sent: Friday, June 19, 2015 5:17 AM
> > > > > To: DECRAENE Bruno IMT/OLN; Stefano Previdi (sprevidi)
> > > > > Cc: spring@ietf.org; isis-wg@ietf.org list
> > > > > Subject: Re: [spring] Conflicting MS entries
> > > > >
> > > > > Even if choosing any IP to MPLS entry does not break anything, 
> > > > > I'm not sure this is a good idea from an operational point of 
> > > > > view to let it
> > > > undeterministic.
> > > > >
> > > > >
> > > > > -----Original Message-----
> > > > > From: spring [mailto:spring-bounces@ietf.org] On Behalf Of 
> > > > > bruno.decraene@orange.com
> > > > > Sent: Thursday, June 18, 2015 09:29
> > > > > To: LITKOWSKI Stephane SCE/IBNF; Stefano Previdi (sprevidi)
> > > > > Cc: spring@ietf.org; isis-wg@ietf.org list
> > > > > Subject: Re: [spring] Conflicting MS entries
> > > > >
> > > > > Hi Stéphane,
> > > > >
> > > > > > From: stephane.litkowski@orange.com> Sent: Thursday, June 
> > > > > > 18,
> > > > > > 2015
> > > > > > 9:23 AM
> > > > > >
> > > > > > Hi Bruno,
> > > > > >
> > > > > > "	 1) I don't really the issue. From a forwarding standpoint, looks
> > like
> > > > > > we can simply program multiple SIDs in the FIB."
> > > > > >
> > > > > > [SLI] What about the IP to MPLS entry ?
> > > > >
> > > > > [Bruno] If transit LSRs install all SIDs, an ingress may use 
> > > > > any SID, no? Local decision.
> > > > >
> > > > > Bruno
> > > > >
> > > > >
> > > > >
> > > > >
> > > >
> > ____________________________________________________________
> > > > ________
> > >
> > >
> > >
> > ____________________________________________________________
> > ________
> > > _____________________________________________________
> > >
> > > Ce message et ses pieces jointes peuvent contenir des informations 
> > > confidentielles ou privilegiees et ne doivent donc pas etre 
> > > diffuses, exploites ou copies sans autorisation. Si vous avez recu 
> > > ce message par erreur, veuillez le signaler a l'expediteur et le 
> > > detruire ainsi que les pieces jointes. Les messages electroniques 
> > > etant susceptibles d'alteration, Orange decline toute 
> > > responsabilite si ce message a ete altere,
> > deforme ou falsifie. Merci.
> > >
> > > This message and its attachments may contain confidential or 
> > > privileged information that may be protected by law; they should 
> > > not be distributed, used or copied without authorisation.
> > > If you have received this email in error, please notify the sender 
> > > and delete this message and its attachments.
> > > As emails may be altered, Orange is not liable for messages that 
> > > have been modified, changed or falsified.
> > > Thank you.
> > >
> > > _______________________________________________
> > > spring mailing list
> > > spring@ietf.org
> > > https://www.ietf.org/mailman/listinfo/spring
> 
> ____________________________________________________________________
> _____________________________________________________
> 
> Ce message et ses pieces jointes peuvent contenir des informations 
> confidentielles ou privilegiees et ne doivent donc pas etre diffuses, 
> exploites ou copies sans autorisation. Si vous avez recu ce message 
> par erreur, veuillez le signaler a l'expediteur et le detruire ainsi 
> que les pieces jointes. Les messages electroniques etant susceptibles 
> d'alteration, Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.
> 
> This message and its attachments may contain confidential or 
> privileged information that may be protected by law; they should not 
> be distributed, used or copied without authorisation.
> If you have received this email in error, please notify the sender and 
> delete this message and its attachments.
> As emails may be altered, Orange is not liable for messages that have 
> been modified, changed or falsified.
> Thank you.
> 
> _______________________________________________
> spring mailing list
> spring@ietf.org
> https://www.ietf.org/mailman/listinfo/spring

_______________________________________________
spring mailing list
spring@ietf.org
https://www.ietf.org/mailman/listinfo/spring

_________________________________________________________________________________________________________________________

Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc
pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler
a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration,
Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.

This message and its attachments may contain confidential or privileged information that may be protected by law;
they should not be distributed, used or copied without authorisation.
If you have received this email in error, please notify the sender and delete this message and its attachments.
As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified.
Thank you.