Hi, Robert:

 

From: [email protected] <[email protected]> On Behalf Of Robert Raszuk
Sent: Wednesday, January 5, 2022 7:55 AM
To: Aijun Wang <[email protected]>
Cc: Greg Mirsky <[email protected]>; Les Ginsberg (ginsberg) 
<[email protected]>; Christian Hopps <[email protected]>; Shraddha Hegde 
<[email protected]>; Tony Li <[email protected]>; Hannes Gredler 
<[email protected]>; lsr <[email protected]>; Peter Psenak <[email protected]>
Subject: Re: [Lsr] BGP vs PUA/PULSE

 

Hi Aijun,

 

In most deployments summary routes are already crafted carefully to only cover 
those destinations which are important and should be reachable from outside of 
the area. 

 

So I see no point in building yet another policy to select a subset of 
summaries to be PUA eligible. 

 

Along those lines I do not buy into the notion of "some prefixes within 
summaries to be more important than others" - it is simply impossible to say 
that this PE is more important than the other one in all practical cases. 

 

If we are to roll out a mechanism to signal unreachability it better be robust 
and dependable - not just an optional hint. With that I would welcome solution 
which says - if we have more then X prefixes (or % of prefixes) to be 
advertised by ABR we stop advertising the summary covering them (or deaggregate 
the summary). 

[WAJ] Actually, 
https://datatracker.ietf.org/doc/html/draft-wang-lsr-prefix-unreachable-annoucement-08#section-7
 has proposed similar solutions, although it should be revised to reflect the 
concerns in this thread and give more careful control for the advertisement of 
PUA/PULSE information. 

 

“Considering the balances of reachable information and unreachable

   information announcement capabilities, the implementation of this

   mechanism should set one MAX_Address_Announcement (MAA) threshold

   value that can be configurable.  Then, the ABR should make the

   following decisions to announce the prefixes:

 

   1.  If the number of unreachable prefixes is less than MAA, the ABR

   should advertise the summary address and the PUAM.

 

   2.  If the number of reachable address is less than MAA, the ABR

   should advertise the detail reachable address only.

 

   3.  If the number of reachable prefixes and unreachable prefixes

   exceed MAA, then advertise the summary address with MAX metric.”

 

 

 

Welcome more idea to address such concerns in more reasonable 

 

 

Thx,

R.

 

 

 

 

 

 

On Wed, Jan 5, 2022 at 12:06 AM Aijun Wang <[email protected] 
<mailto:[email protected]> > wrote:

I think the mentioned solution can also address Robert and Christian’s concerns.

Aijun Wang

China Telecom





On Jan 5, 2022, at 07:02, Aijun Wang <[email protected] 
<mailto:[email protected]> > wrote:

Hi, Greg:

 

https://datatracker.ietf.org/doc/html/draft-wang-lsr-prefix-unreachable-annoucement-08#section-8
 has some description for such considerations: 

“ In order to reduce the unnecessary advertisements of PUAM

   messages on ABRs, the ABRs should support the configuration of the
   protected prefixes.  Based on such information, the ABR will only
   advertise the PUAM message when the protected prefixes(for example,
   the loopback addresses of PEs that run BGP) that within the summary
   address is missing.”

 
 

Aijun Wang

China Telecom





On Jan 5, 2022, at 03:56, Greg Mirsky <[email protected] 
<mailto:[email protected]> > wrote:



Hi Peter,

thank you for your response. I'm looking forward to the new version of the 
draft. It will be interesting to learn the criteria that enable an ABR to 
reliably identify the scenarios you've suggested are outside the scope of the 
PULSE draft and should be handled differently.

 

Regards,

Greg

 

On Tue, Jan 4, 2022 at 10:08 AM Peter Psenak <[email protected] 
<mailto:[email protected]> > wrote:

Hi Greg,

On 04/01/2022 18:13, Greg Mirsky wrote:
> Hi Peter,
> I'm probably missing something in the current PULSE but I cannot find 
> the mechanism that limits the number of the pulses. Do you envision that 
> being like a throttling mechanism? But delaying the propagation of 
> notification for some events might cause more instability in a network.

no. It's a limit not a delay. If too many edge nodes loose connectivity 
to the ABR in its area, it's a result of the severe event like area 
partition or loss of area connectivity from ABR. These are not types of 
events that we are trying to address with pulses.

The limit is not described in the published version of the draft.
We are working on the updated version that will include the description 
of it.

thanks,
Peter

> 
> Regards,
> Greg
> 
> On Tue, Jan 4, 2022 at 1:52 AM Peter Psenak <[email protected] 
> <mailto:[email protected]>  
> <mailto:[email protected] <mailto:[email protected]> >> wrote:
> 
>     Hi Greg,
> 
>     On 03/01/2022 23:17, Greg Mirsky wrote:
>      > Happy New Year to All!
>      >
>      > Hi Peter,
>      > Top-pasting:
>      > In 99,99% of cases there will be only single pulse generated when
>     one PE
>      > goes down. That itself is a very rare event itself.
>      >
>      > We can easily limit the number of pulses generated on ABR to a single
>      > digit number to cover the unlikely case of many PEs in area becoming
>      > unreachable at the same time.
>      >
>      > I think that it is possible for the summarizing ABR to get
>     disconnected
>      > from the IGP area in such a way that the summarization is still
>     valid.
>      > If such a case is valid, would the ABR generate PULSE for each
>      > disconnected PE?
> 
>     obviously not. That's why I mentioned the number of pulses will be
>     limited on every ABR.
> 
>     thanks,
>     Peter
> 
>      >
>      > Regards,
>      > Greg
>      >
>      > On Mon, Jan 3, 2022 at 8:56 AM Peter Psenak
>      > <[email protected] 
> <mailto:[email protected]> 
>     <mailto:[email protected] <mailto:[email protected]> >
>     <mailto:[email protected] <mailto:[email protected]> 
>     <mailto:[email protected] <mailto:[email protected]> >>>
>      > wrote:
>      >
>      >     Chris,
>      >
>      >     On 03/01/2022 17:18, Christian Hopps wrote:
>      >      >
>      >      > Peter Psenak <[email protected] <mailto:[email protected]>  
> <mailto:[email protected] <mailto:[email protected]> >
>     <mailto:[email protected] <mailto:[email protected]>  
> <mailto:[email protected] <mailto:[email protected]> >>> writes:
>      >      >
>      >      >> On 03/01/2022 16:21, Christian Hopps wrote:
>      >      >>>
>      >      >>>> On Nov 29, 2021, at 7:39 PM, Les Ginsberg (ginsberg)
>      >     <[email protected] 
> <mailto:[email protected]> 
>     <mailto:[email protected] <mailto:[email protected]> >
>      >     <mailto:[email protected] 
> <mailto:[email protected]> 
>     <mailto:[email protected] <mailto:[email protected]> 
> >>> wrote:
>      >      >>>>
>      >      >>>> Tony –
>      >      >>>>    Let me try one example – see if it helps.
>      >      >>>>    Summarization is used in the network.
>      >      >>>> But customer identifies a modest number of key nodes
>     where it
>      >     wants to detect loss of reachability ASAP. Unfortunately,
>     customer
>      >     is unable to assign addresses which are outside of the summary to
>      >     these nodes.
>      >      >>>
>      >      >>> I think this does in fact capture the problem trying to be
>      >     solved here, nicely.
>      >      >>
>      >      >> not really.
>      >      >> In fact assigning addresses to the nodes in a way that
>     they are
>      >     part of the
>      >      >> summary is the right thing to do.
>      >      >
>      >      > No, not if you want more detailed information about specific
>      >     reachability it's not. And ....
>      >
>      >
>      >     typically you want to summarize all prefixes inside the area when
>      >     advertising outside the area. And you want to know about some
>     of these
>      >     prefixes when they are lost to help convergence.
>      >
>      >
>      >      >
>      >      >> The problem we are trying to solve is to use the
>     summarization
>      >     but without the
>      >      >> loss of the fast notification of the node down event.
>      >      >
>      >      > You want more specific information about reachability, but you
>      >     just want to do it when the network is stressed and
>     undergoing change.
>      >      >
>      >      > So the "works now" way of not summarizing these important
>      >     prefixes has the state in the network when it's working, so
>     you know
>      >     adding and removing it is something the network is already
>     capable
>      >     of handling.
>      >      >
>      >      > New signaling that *only* is created when things start
>     failing,
>      >     tests the infrastructure at exactly the wrong time.
>      >
>      >     In 99,99% of cases there will be only single pulse generated when
>      >     one PE
>      >     goes down. That itself is a very rare event itself.
>      >
>      >     We can easily limit the number of pulses generated on ABR to
>     a single
>      >     digit number to cover the unlikely case of many PEs in area
>     becoming
>      >     unreachable at the same time.
>      >
>      >
>      >      >
>      >      > If a failing network can handle the extra state, then a
>      >     functioning stable network of course can too.
>      >
>      >     no, that's not what we claim. We want network to be
>     summarized all
>      >     times
>      >     and generate limited number of pulses at any given time to
>     help the
>      >     network converge fast in case where single (or very few) PEs
>     in an area
>      >     go down.
>      >
>      >     thanks,
>      >     Peter
>      >
>      >
>      >
>      >      >
>      >      > Thanks,
>      >      > Chris.
>      >      >
>      >      >>
>      >      >> thanks,
>      >      >> Peter
>      >      >>
>      >      >>
>      >      >>> One solution very simple solution that works today is:
>      >      >>> - Tell the customer they can't do this, but they *can*
>     modify
>      >     their addressing
>      >      >>> (this is literally what they do for a living) so that they
>      >     don't have this
>      >      >>> problem.
>      >      >>> Do we *really* want modify our IGPs (a BIG ask) with some
>      >     pretty questionable
>      >      >>> changes, just to save the operators the trouble of doing
>     their
>      >     job correctly?
>      >      >>> Maybe the answer here is this isn't a good idea, and we
>     should
>      >     move on...
>      >      >>> Thanks,
>      >      >>> Chris.
>      >      >>> [as wg member]
>      >      >>> _______________________________________________
>      >      >>> Lsr mailing list
>      >      >>> [email protected] <mailto:[email protected]>  <mailto:[email protected] 
> <mailto:[email protected]> > <mailto:[email protected] <mailto:[email protected]> 
>     <mailto:[email protected] <mailto:[email protected]> >>
>      >      >>> https://www.ietf.org/mailman/listinfo/lsr
>     <https://www.ietf.org/mailman/listinfo/lsr>
>      >     <https://www.ietf.org/mailman/listinfo/lsr
>     <https://www.ietf.org/mailman/listinfo/lsr>>
>      >      >>>
>      >      >
>      >      >
>      >
>      >     _______________________________________________
>      >     Lsr mailing list
>      > [email protected] <mailto:[email protected]>  <mailto:[email protected] 
> <mailto:[email protected]> > <mailto:[email protected] <mailto:[email protected]> 
>     <mailto:[email protected] <mailto:[email protected]> >>
>      > https://www.ietf.org/mailman/listinfo/lsr
>     <https://www.ietf.org/mailman/listinfo/lsr>
>      >     <https://www.ietf.org/mailman/listinfo/lsr
>     <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]
https://www.ietf.org/mailman/listinfo/lsr

Reply via email to