Re: [Lsr] Working Group Last Call for draft-ietf-lsr-ip-flexalgo-04 - "IGP Flexible Algorithms (Flex-Algorithm) In IP Networks"

["Les Ginsberg (ginsberg)" <[e-mail]>]

Aijun –

I am not an author of the draft – and so cannot speak on behalf of the draft authors.
But here is my response as WG member.

You need to focus on the dataplane.

Suppose a node advertises 1.1.1.1/32 in (IS-IS) TLV 135.
If a packet addressed to 1.1.1.1 arrives unlabeled, it can be forwarded using the Algo 0 path(s) installed in forwarding.
If a packet addressed to 1.1.1.1 arrives with MPLS label encap, you can use the algorithm specific SID to do a lookup and forward based on the flex-algo specific paths.

But if you do not have an SR encap, what about the packet will tell you to forward via (for example) Flex-algo 130 paths rather than Algo 0 paths? And how would you install such paths in the dataplane along with Algo 0 paths?

IP Flex-Algo addresses those issues by assigning a given address to one-and-only-one algo. This is why new Reachability advertisements are required.
So, for example, the same node could advertise 1.1.1.2/32 in the new IPv4 Algorithm Prefix Reachability TLV, associate it with algorithm 130, and a forwarding entry can be installed for that prefix that follows Algo 130 paths.
Since this address is not used by any other algorithm (and especially not by Algo 0), there is no ambiguity in the dataplane.

This is also why the statement you quote below regarding the same prefix advertised in both IPv4 Reachability and IPv4 Algorithm Prefix Reachability is necessary. The dataplane cannot support both for the same prefix.

HTH

    Les


From: Aijun Wang <[e-mail]>
Sent: Tuesday, May 3, 2022 3:10 PM
To: Les Ginsberg (ginsberg) <[e-mail]>
Cc: Peter Psenak (ppsenak) <[e-mail]>; Acee Lindem (acee) <[e-mail]>; [e-mail]; [e-mail]
Subject: Re: [Lsr] Working Group Last Call for draft-ietf-lsr-ip-flexalgo-04 - "IGP Flexible Algorithms (Flex-Algorithm) In IP Networks"

Hi, Peter and Les:
Prefix Segment Identifier sub-TLV and FAPM sub-TLV are two independent sub-TLVs for TLV135, 235,236 and 237.
They are not required to be exist at the same time.
FAPM just describes the metrics that associated with different Flex-Algo. Isn’t it more straightforward to associate it with the intended prefixes to achieve the same results as the newly defined top-TLV?
I want to know the technical reason why we can’t follow this direction?
Anyway, the router can use such information to calculate the different forwarding path based on the algorithm and metric.
If there is any obstacles to achieve this, I think we should update the https://datatracker.ietf.org/doc/html/draft-ietf-lsr-flex-algo-19 to cover it.

And, in the WGLC document, there is the following description:


  In cases where a prefix advertisement is received in both a IPv4

   Prefix Reachability TLV and an IPv4 Algorithm Prefix Reachability

   TLV, the IPv4 Prefix Reachability advertisement MUST be preferred

   when installing entries in the forwarding plane.

It is obvious that any prefixes can be advertised in either TLVs, what’s the necessary to define the new one?


Aijun Wang
China Telecom


On May 3, 2022, at 22:48, Les Ginsberg (ginsberg) <ginsberg=[e-mail]<mailto:ginsberg=[e-mail]>> wrote:


Peter -



I am in agreement.



However, the IANA section of the draft is missing some necessary information.

The new top level TLVs in IS-IS - I am assuming you want these to share the sub-TLV space defined in https://www.iana.org/assignments/isis-tlv-codepoints/isis-tlv-codepoints.xhtml#isis-tlv-codepoints-advertising-prefix-reachability

In which case you need to provide a list of the existing sub-TLVs and an indication (Y/N) as to whether they are allowed in the new TLVs.



Here is my initial take:



1    32-bit Administrative Tag Sub-TLV    Y

2    64-bit Administrative Tag Sub-TLV    y

3    Prefix Segment Identifier            n

4    Prefix Attribute Flags              y

5    SRv6 End SID                        n

6    Flex-Algorithm Prefix Metric        n

11   IPv4 Source Router ID               y

12   IPv6 Source Router ID               y

32   BIER Info                          n(??)



    Les



> -----Original Message-----

> From: Lsr <[e-mail]<mailto:[e-mail]>> On Behalf Of Peter Psenak

> Sent: Tuesday, May 3, 2022 7:14 AM

> To: Aijun Wang <[e-mail]<mailto:[e-mail]>>; Peter Psenak

> <ppsenak=[e-mail]<mailto:ppsenak=[e-mail]>>

> Cc: Acee Lindem (acee) <acee=[e-mail]<mailto:acee=[e-mail]>>; [e-mail]<mailto:[e-mail]>;

> [e-mail]<mailto:[e-mail]>

> Subject: Re: [Lsr] Working Group Last Call for draft-ietf-lsr-ip-flexalgo-04 -

> "IGP Flexible Algorithms (Flex-Algorithm) In IP Networks"

>

> Aijun,

>

> On 03/05/2022 15:52, Aijun Wang wrote:

> > Hi, Peter:

> > I think the logic is the following:

> > FAPM is the sub-TLV of TLV 135,235,236 and 237, then it extends these TLVs

> for advertising prefixes in algorithm 0 to other Flexible Algorithm.

> > Then I see no reason to define the new top-TLV to encoding the similar

> information.

>

> FAPM is used in SR-MPLS case where algo 0 prefix has multiple flex-algo

> SIDs. So the algo 0 reachability is always advertised in legacy TLV and

> FAPM is used to advertise additional flex-algo metric for inter-area or

> external prefixes.

>

> We can not use it for IP flex-algo.

>

> thanks,

> Peter

>

> >

> > Aijun Wang

> > China Telecom

> >

> >> On May 3, 2022, at 19:16, Peter Psenak

> <ppsenak=[e-mail]<mailto:ppsenak=[e-mail]>> wrote:

> >>

> >> Hi Aijun,

> >>

> >>> On 03/05/2022 11:57, Aijun Wang wrote:

> >>> Hi, Peter：

> >>> Different data planes use different Flex-Algorithm and associated

> metric, they can’t be mixed.

> >>> Or, would you like to point out why the following scenarios can’t be

> achieved via the FAPM?

> >>> 1) The PE router has three loopback addresses(Lo1-Lo3), each associated

> with different Flex-ALgorithhms, and also different metrics. They are

> advertised via the FAPM, no MPLS SIDs are associated with these loopack

> prefixes advertisements.

> >>> 2) The PE router has also another inter-area/inter-domain

> prefixes(IPextra), with the FAPM and MPLS SID advertised via the prefixes

> advertisements.

> >>> When the PE in other ends want to send the traffic to theses addresses:

> >>> 1)  To the formers three destinations(Lo1-Lo3), the FIB that are formed

> by the associated FAPM will be used, that is, the IP-based forwarding will be

> selected.

> >>> 2) To the Inter-area/inter-domain prefixes the FIB that are formed via

> the FAPM and the associated SID, the MPLS-based forwarding will be

> selected.

> >>> Why can’t they coexist?

> >>

> >> FAPM Sub-TLV is a sub-TLV of TLVs 135, 235, 236, and 237. These TLVs

> advertise the reachability of the prefix in algorithm 0.

> >>

> >> For an IP algo prefix, which is associated with the flex-algorithm, the

> reachability in algorithm 0 must not be advertised. So we have to use a

> different top level TLV.

> >>

> >>

> >> thanks,

> >> Peter

> >>

> >>

> >>

> >>> Aijun Wang

> >>> China Telecom

> >>>>> On May 3, 2022, at 16:05, Peter Psenak

> <ppsenak=[e-mail]<mailto:ppsenak=[e-mail]>> wrote:

> >>>>

> >>>> Aijun,

> >>>>

> >>>>> On 03/05/2022 09:59, Aijun Wang wrote:

> >>>>> Hi, Peter:

> >>>>> The definition of FAPM for IS-IS and OSPF doesn’t prevent from it is

> used for the intra-area prefixes.

> >>>>> If we advertise the different loopback addresses via the FAPM,

> associate them to different Flex-Algo and related metrics, and does not

> allocate the MPLS SID, we can achieve the IP-Flex effect then.

> >>>>

> >>>> as I said, we can not mix metrics for different data-planes.

> >>>>

> >>>>> So, what’s the additional value of the IP-Flexalgo draft then?

> >>>>

> >>>> please read the draft. It defines the flex-algo for IP data plane.

> >>>>

> >>>> thanks,

> >>>> Peter

> >>>>

> >>>>

> >>>>

> >>>>> Aijun Wang

> >>>>> China Telecom

> >>>>>>> On May 3, 2022, at 14:46, Peter Psenak

> <ppsenak=[e-mail]<mailto:ppsenak=[e-mail]>> wrote:

> >>>>>>

> >>>>>> Aijun,

> >>>>>>

> >>>>>>> On 03/05/2022 00:47, Aijun Wang wrote:

> >>>>>>> Hi, Acee:

> >>>>>>> The questions raised at

> https://mailarchive.ietf.org/arch/msg/lsr/RlHphXCwxMbgGvcBV_m24xiDzS0<https://mailarchive.ietf.org/arch/msg/lsr/RlHphXCwxMbgGvcBV_m24xiDzS0/>

> /<https://mailarchive.ietf.org/arch/msg/lsr/RlHphXCwxMbgGvcBV_m24xiDzS0/>

> <https://mailarchive.ietf.org/arch/msg/lsr/RlHphXCwxMbgGvcBV_m24xiDzS<https://mailarchive.ietf.org/arch/msg/lsr/RlHphXCwxMbgGvcBV_m24xiDzS0/>

> 0/<https://mailarchive.ietf.org/arch/msg/lsr/RlHphXCwxMbgGvcBV_m24xiDzS0/>> has not been answered.

> >>>>>>

> >>>>>> IS-IS Flexible Algorithm Prefix Metric Sub-TLV” and “OSPF Flexible

> Algorithm Prefix Metric Sub-TLV” are defined for advertisement of algorithm

> specific metric for inter-area inter-AS prefixes for SR-MPLS data-plane.

> >>>>>>

> >>>>>> SR MPLS and IP are independent data-planes used for flex-algo. We

> can not mix their metric.

> >>>>>>

> >>>>>> thanks,

> >>>>>> Peter

> >>>>>>

> >>>>>>> Aijun Wang

> >>>>>>> China Telecom

> >>>>>>>>> On May 2, 2022, at 23:00, Acee Lindem (acee)

> <acee=[e-mail]<mailto:acee=[e-mail]>> wrote:

> >>>>>>>>

> >>>>>>>> The WG last call has completed. We will submit an updated

> version of the document for publication with the terminology changes based

> on the discussion amongst the authors, Ketan, Robert, Gyan, and others.

> >>>>>>>>

> >>>>>>>> Thanks,

> >>>>>>>> Acee

> >>>>>>>>

> >>>>>>>> *From: *Lsr <[e-mail]<mailto:[e-mail]>> on behalf of "Acee Lindem

> (acee)" <acee=[e-mail]<mailto:acee=[e-mail]>>

> >>>>>>>> *Date: *Thursday, April 7, 2022 at 3:07 PM

> >>>>>>>> *To: *"[e-mail]<mailto:[e-mail]>" <[e-mail]<mailto:[e-mail]>>

> >>>>>>>> *Cc: *"[e-mail]<mailto:[e-mail]>" <draft-ietf-lsr-ip-<mailto:[e-mail]>

> [e-mail]<mailto:[e-mail]>>

> >>>>>>>> *Subject: *[Lsr] Working Group Last Call for draft-ietf-lsr-ip-

> flexalgo-04 - "IGP Flexible Algorithms (Flex-Algorithm) In IP Networks"

> >>>>>>>>

> >>>>>>>> This begins a WG last call for draft-ietf-lsr-ip-flexalgo-04.  The draft

> had a lot of support and discussion initially and has been stable for some

> time. Please review and send your comments, support, or objection to this

> list before 12 AM UTC on April 22^nd , 2022.

> >>>>>>>>

> >>>>>>>> Thanks,

> >>>>>>>> Acee

> >>>>>>>>

> >>>>>>>> _______________________________________________

> >>>>>>>> Lsr mailing list

> >>>>>>>> [e-mail]<mailto:[e-mail]>

> >>>>>>>> https://www.ietf.org/mailman/listinfo/lsr

> >>>>>>

> >>>>

> >>> _______________________________________________

> >>> Lsr mailing list

> >>> [e-mail]<mailto:[e-mail]>

> >>> https://www.ietf.org/mailman/listinfo/lsr

> >>

> >

> >

>

> _______________________________________________

> Lsr mailing list

> [e-mail]<mailto:[e-mail]>

> https://www.ietf.org/mailman/listinfo/lsr
_______________________________________________
Lsr mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/lsr