Re: [Lsr] Flooding across a network

[[e-mail]]

Hello Gyan,

Nice to know that fast LSP flooding is important to you.
You are right that there are existing code and knobs to improve IGP convergence in general, and in particular to the specific implementation you are referring to (I’d bet that its first letter start with C).
Focusing on efficient flooding, which is the subject of this thread, “fast-flood x” is indeed a nice name. AFAIR, the feature has been introduced approximately 15 years ago and was indeed an improvement. Many more have been introduced as part of the fast IGP convergence project that I’d been running in this company, resulting in very significant and much needed improvement on that implementation.
Despite the name “fast flood”, I don’t think that anyone is claiming that it’s solving the broad issue of efficient LSP flooding.
LSP-pacing may also be configured in order to influence the delay between two consecutively sent LSP.

One question, which you have likely faced when enabling these knobs, is the right values to configure, depending on the receiver (’s capability).
If you have multiple implementation on the sender side –with different ways to implement ‘fast flooding’- and different platforms (‘s capability) on the receiver side, combined with various usages (e.g. from 2 to 50 IGP adjacencies, with or without BGP/RSVP-TE/multicast….) the matrix is quickly getting significant. Your vendor of the receiver implementation may help you; or not (significantly).
With IS-IS, the problem is exacerbated because the protocol is typically vital to the network and all its usages and customers. So nobody wants to be held accountable if the parameters turned to be ‘non-optimal’ over the next decade. Especially as routing ‘incidents’ may already happen with default configuration.


Sending fast is “easy”, but what we need is an effective communication between the two entities involved: one sender and a receiver; sometimes referred to as goodput. RFC 2914 “Congestion Control Principles” [1] may be an introduction to this.
e.g. you can probably speak English to me at 200-250 words per minutes but I can assure you that it would be more effective to speak to me at a lower rate, because of limitations on the receiver side.
That’s similar for machines in general and IS-IS in particular, especially since IS-IS is not that fast to recover from lost LSPs.
The problem is not new and referred to as flow control. From [2] “In data communications<https://en.wikipedia.org/wiki/Data_communications>, flow control is the process of managing the rate of data transmission between two nodes to prevent a fast sender from overwhelming a slow receiver. It provides a mechanism for the receiver to control the transmission speed, so that the receiving node is not overwhelmed with data from transmitting node.”

That’s what we are working on.

Kind regards,
Bruno

[1] https://tools.ietf.org/html/rfc2914
[2] https://en.wikipedia.org/wiki/Flow_control_(data)



From: Gyan Mishra [mailto:[e-mail]]
Sent: Sunday, May 17, 2020 11:07 AM
To: DECRAENE Bruno TGI/OLN
Cc: Christian Hopps; Les Ginsberg (ginsberg); [e-mail]
Subject: Re: [Lsr] Flooding across a network


Am reading through this thread late but want to chime in on the discussion.

Router Isis
fast-flood x

The concept of ISIS fast flooding has been around for decades and is critical for ISIS LSDB synchronization to so all nodes have all prefixes to avoid micro loops and within the network.

The concept of fast flood feature is basically to flood the number of LSPs that need to be flooded before starting the local SPF.  So with fast flood enabled all nodes in the domain wait till the last LSP is flooded before it runs its local SPF.  By fast flooding it reduces the overall number of SPFs executed and allows the LSDB to be synchronized by all nodes in the domain.

Let’s say a link went down or came up topology change.  The router should at a minimum flood at least that LSP that triggered the SPF before running its local SPF.  Fast flooding is recommended by all vendors as it improves convergence times and limits duration of LSDB inconsistencies during which time  micro loops form.

In case where a few routers don’t have fast flood enabled, those  few router in slow mode may have to run multiple SPFs before completely being synchronized.  I think that question where a mix of slow and fast exist versus all slow or all fast you may have mixed unpredictable behaviors that you really would have to test out in a lab or live network.

In my experience I have always had all nodes running fast flood mode as the feature has been around for a long time and recommended for optimal convergence.

There are many other parameters with ISIS that come into play to optimize convergence below which may vary by vendor but can improve convergence and minimize micro loops below:

One of them is setting all P2P routed links to circuit type point-to-point to avoid DIS election.

Both ISIS and OSPF have the iSPF incremental SPF so only the changed part of the tree is updated and not the entire tree to save on SPF processing.

With short spf interval and spf delay the local spf May start before the lsp that triggered the spf is flooded.

Lsp pacing can be used to speed up end to end flooding.

Hello padding TLV 8 for MTU defection to help with convergence so max number of LSPs can be sent during flooding.

Increase lsp lifetime to maximum to reduce control traffic so cpu can be optimized to refresh lsp’s

Reduce the frequency of periodic lsp flooding of topology to reduce link utilization by Isis

Ignore lsp errors instead of purging them

Carful balance of event processing and throttling of events to maximize convergence times

Prefix prioritization on critical prefixes such as MPLS FEC binding.

LFA and RLFA for pre computed backup paths

BFD for ms failure detection

Overall Isis as far as scalability far exceeds ospf with stability with domains with a very large number of nodes in the 100s and high number of adjacencies per node.

As far as stability and convergence both ospf and Isis are equivalent from my experience.

Kind regards

Gyan
Verizon

On Thu, May 7, 2020 at 12:03 PM <[e-mail]<mailto:[e-mail]>> wrote:
Les,


From: Les Ginsberg (ginsberg) [mailto:[e-mail]<mailto:[e-mail]>]
Sent: Thursday, May 7, 2020 4:55 PM
To: DECRAENE Bruno TGI/OLN
Cc: [e-mail]<mailto:[e-mail]>; Christian Hopps
Subject: RE: [Lsr] Flooding across a network

Bruno –

I have specifically used an example where “microloop avoidance” is not applicable. So I did not want to use the term “microloop” but rather used “loop” so as not to suggest that “microloop avoidance” is a potential solution for the sub-optimal behavior.
[Bruno] I’m not sure what you mean exactly by “microloop avoidance”.
On my side, I mean “Loop avoidance using Segment Routing” [1] which _is_ applicable. (Note that I’m not saying that all _implementations_ cover all cases.)
[1]  https://tools.ietf.org/html/draft-bashandy-rtgwg-segment-routing-uloop-08


Hope you can appreciate that point.

It would be easy enough to include more nodes in the topology which only support slow flooding. The end result would be the same.
[Bruno] The question is whether the activation of fast flooding on one (some) node(s)/IGP adjacency(es) may result in delaying the LSDB synchronization network wide. It’s not about the number of slow nodes.

You seem to assume that we have a majority of fast nodes, and some remaining slow nodes. In which cases I agree that in some cases the overall/network wide behavior may be a slow LSDB sync. But not slower than with slow nodes only.
I have kept the example simple in the hopes we could more easily agree that what I describe can happen when not all nodes support faster flooding – which is the only point I am trying to make.
[Bruno] If your point is that while we have slow nodes in the network, in some cases the network wide behavior can be “slow”, somewhere between “all nodes are fast” and “all nodes are slow”, then I agree with you.

However what you said is: “when only some nodes in the network support faster flooding […]  it prolongs the period of LSDB inconsistency.”
In that sentence, I disagree with “prolongs” which according to 2 dictionaries means “lasting longer”.
https://dictionary.cambridge.org/fr/dictionnaire/anglais/prolong
https://www.merriam-webster.com/dictionary/prolong

Because, at least to me, this reads as “The introduction of fast-flooding nodes in the network (may) increase the period of LSDB inconsistency.”

Whether the ratio of Fast/Slow nodes is large or small or about the same doesn’t eliminate the possibility that the same behavior could be seen – though it might alter the location of the topology change which would be problematic.

From an operator’s POV, I am pretty sure that what you really care about is whether packets get successfully forwarded or not.
[Bruno] Yes, but let’s not combine multiple (complex) problems.
Here I’d like we focus on LSP flooding and LSDB synchronization across the network. Plus this is exactly the point that you raised: “increase the period of LSDB inconsistency.”

I am demonstrating that it isn’t safe to assume forwarding behavior will be optimal when not all nodes support fast flooding.
[Bruno] Which is very different from your original point “when only some nodes in the network support faster flooding […]  it prolongs the period of LSDB inconsistency.”

--Bruno

   Les


From: [e-mail]<mailto:[e-mail]> <[e-mail]<mailto:[e-mail]>>
Sent: Thursday, May 07, 2020 2:18 AM
To: Les Ginsberg (ginsberg) <[e-mail]<mailto:[e-mail]>>
Cc: [e-mail]<mailto:[e-mail]>; Christian Hopps <[e-mail]<mailto:[e-mail]>>
Subject: RE: [Lsr] Flooding across a network


Les,



> From: Les Ginsberg (ginsberg) [mailto:[e-mail]]

>

> Bruno -

>

> I am sorry it has been so difficult for us to understand each other. I am trying my best.

+1 +1



> Look at it this way:

>

> You are the customer. 😊

> I am the vendor.



I'm not sure what (technical) point you are trying to make.



Coming back to your statement : “when only some nodes in the network support faster flooding […]  it prolongs the period of LSDB inconsistency.”

Your example is flawed on two points:

- Your performance indicator is "micro loop duration", while we are talking about the duration of LSP flooding across the network. So the metric should be "LSP flooding time" (or "period of LSDB inconsistency")

- Your example is about one node doing slower flooding, while we are interested in the case when one node support faster flooding. (It's quite clear that if one node is slower, the end to end flooding time may be longer. The point you are raising is when one node is doing faster flooding)



Can you fix your example on those 2 points?



Thanks

--Bruno





> The failure scenario I describe below happens and you notice that all Northbound destinations loop for 35 seconds whenever fast flooding is enabled.

> I think you are going to complain about this - to me. 😊

>

> And I am going to tell you that this is a consequence of enabling fast flooding in the presence of a node which does not support it. Your options to reduce the period of looping will be:

>

> 1)Upgrade the slow node to support faster flooding

> 2)Disable fast flooding

> 3)Redesign your network

>

    > Les

>

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

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

> > Sent: Wednesday, May 06, 2020 10:10 AM

> > To: Christian Hopps <[e-mail]<mailto:[e-mail]>>

> > Cc: Les Ginsberg (ginsberg) <[e-mail]<mailto:[e-mail]>>; [e-mail]<mailto:[e-mail]>

> > Subject: RE: [Lsr] Flooding across a network

> >

> > > From: Christian Hopps [mailto:[e-mail]]

> > >

> > > Bruno persistence has made me realize something fundamental here.

> > >

> > > The minute the LSP originator changes the LSP and floods it you have LSDB

> > inconsistency.

> >

> > Exactly my point. Thank you Chris.

> > I would even say: "The minute the LSP originator changes the LSP then you

> > have LSDB inconsistency." But no big deal if there is disagreement on this

> > detail.

> >

> > > That is going to last until the last node in the network has updated it's LSDB.

> >

> > Absolutely.

> > So the faster we flood, the shorter the LSBD inconsistency.

> >

> > Now IMO, even if a single/few nodes flood faster, there is a chance of

> > shortening the LSDB inconsistency. But in all cases, I don't see how this could

> > make the LSDB inconsistency longer.

> >

> >

> > > Les is pointing out that LSDB inconsistency can be bad in certain

> > circumstances e.g., if a critical node is slow and thus inconsistent.

> > >

> > > I believe the right way to fix this is a simple one, help the operator flag the

> > broken router software/hardware for replacement, but otherwise IS-IS

> > should just try to do the best job it can do to which is to flood around the

> > problem (i.e., flood as optimally as possible).

> >

> > +1

> > On a side note, I would not call a router flooding slowly as "broken". I find it

> > understandable that in a given network there are different type of routers

> > (core vs aggregation), different roles (P having 50 IGP adjacencies with 50 PEs

> > vs PE having only 2 IGP adjacencies with 2 P), different hardware

> > generations, different software, different vendors with different

> > perspectives/markets.

> >

> > Thank you Chris.

> >

> > --Bruno

> > >

> > > Thanks,

> > > Chris.

> > > [as WG member]

> > >

> > >

> > > > On May 6, 2020, at 10:33 AM, [e-mail]<mailto:[e-mail]> wrote:

> > > >

> > > > Les,

> > > >

> > > > From: Les Ginsberg (ginsberg) [mailto:[e-mail]]

> > > > Sent: Wednesday, May 6, 2020 4:14 PM

> > > > To: DECRAENE Bruno TGI/OLN

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

> > > > Subject: RE: Flooding across a network

> > > >

> > > > Bruno –

> > > >

> > > > I am somewhat at a loss to understand your comments.

> > > > The example is straightforward and does not need to consider FIB update

> > time nor the ordering of prefix updates on different nodes.

> > > > [Bruno] The example is straightforward but you are referring to FIB and IP

> > packets forwarding as per those FIBs.

> > > > I’d like we focus on LSP flooding and LSDB consistency.

> > > >

> > > > Consider the state of Node B and Node D at various time points from the

> > trigger event.

> > > >

> > > > T+ 2 seconds:

> > > > -----------------

> > > > B has received all LSP Updates. It triggers an SPF and for all Northbound

> > destinations previously reachable via C it installs paths via D.

> > > > Let’s assume it take 5 seconds to update the forwarding plane.

> > > >

> > > > D has received 40 of the 1000 LSP updates. It triggers an SPF and finds

> > that all Northbound destinations are reachable via B-C. It makes no changes

> > to the forwarding plane.

> > > >

> > > > T+7 seconds

> > > > -----------------

> > > > B has completed FIB updates. Traffic to all Northbound destinations is

> > being forwarded via D.

> > > >

> > > > D has now received 140 of the 1000 LSP updates. Entries in its forwarding

> > plane for Northbound destinations still point to B.

> > > >

> > > > We have a loop.

> > > >

> > > > T + 30 seconds

> > > > --------------------

> > > > D has now received 600 of the 1000 LSP updates. Still no changes to its

> > forwarding plane.

> > > > Traffic to Northbound destinations is still looping.

> > > >

> > > > T+ 50 seconds

> > > > -------------------

> > > > D has finally received all 1000 LSP updates..

> > > > It triggers (another) SPF and calculates paths to Northbound destinations

> > via E. It begins to update its forwarding plane.

> > > > Let’s assume this will take 5 seconds..

> > > >

> > > > T + 55 seconds

> > > > --------------------

> > > > D has completed forwarding plane updates – no more looping.

> > > >

> > > > That is all I am trying to illustrate.

> > > >

> > > > If you want to start arguing that node protecting LFAs + microloop

> > avoidance could help (NOTE I explicitly  took those out of the example for

> > simplicity) – it is easy enough to change the example to include multiple node

> > failures or a node failure plus some northbound link failures on other nodes.

> > > > [Bruno] I’m not talking about LFA/FRR. And with regards to microloops

> > avoidance, some algorithms can handle any graph transition so including

> > multiple node failures.

> > > >

> > > > But again, let’s stick to LSP flooding and LSDB consistency. (you are the

> > one speaking about microloops in the forwarding plane).

> > > >

> > > > The point here is to look at the impact of long-lived LSDB inconsistency

> > which results when some nodes support flooding an order of magnitude

> > faster flooding than other nodes – which is what you asked me to clarify.

> > > > [Bruno] No. I asked you to clarify why having a node with faster flooding

> > could prolongs the period of LSDB inconsistency.

> > > >

> > > > Again, with you own words: “when only some nodes in the network

> > support faster flooding the behavior of the whole network may not be

> > "better" when faster flooding is enabled because it prolongs the period of

> > LSDB inconsistency.”

> > > > And with less words: “when only some nodes in the network support

> > faster flooding […]  it prolongs the period of LSDB inconsistency.”

> > > >

> > > > --Bruno

> > > >

> > > >    Les

> > > >

> > > >

> > > >

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

> > > > Sent: Wednesday, May 06, 2020 6:21 AM

> > > > To: Les Ginsberg (ginsberg) <[e-mail]<mailto:[e-mail]>>

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

> > > > Subject: RE: Flooding across a network

> > > >

> > > > Les,

> > > >

> > > > From: Les Ginsberg (ginsberg) [mailto:[e-mail]]

> > > > Sent: Wednesday, May 6, 2020 1:35 AM

> > > > To: DECRAENE Bruno TGI/OLN; lsr@ietf..org<mailto:lsr@ietf..org>

> > > > Subject: RE: Flooding across a network

> > > >

> > > > Bruno -

> > > >

> > > > Seems like it was not too long ago that we were discussing this in person.

> > Ahhh...the good old days...

> > > > [Bruno] Indeed, may be not to the point of concluding. Indeed.

> > > >

> > > > First, let's agree that the interesting case does not involve 1 or even a

> > small number of LSPs. For those cases flooding speed does not matter.

> > > > The interesting cases involve a large number of LSPs (hundreds or

> > thousands). And in such cases LFA/microloop avoidance techniques are not

> > applicable.

> > > >

> > > > Take the following simple topology:

> > > >

> > > >    |  | ... |            |

> > > >      +---+             +---+

> > > >      | C |             | E |

> > > >      +---+             +---+

> > > >        |                 | 1000

> > > >      +---+             +---+

> > > >      | B |-------------| D |

> > > >      +---+   1000      +---+

> > > >        |                 |

> > > >        |                 |

> > > >         \               /

> > > >          \            /

> > > >           \         /

> > > >            \      /

> > > >              +---+

> > > >              | A |

> > > >              +---+

> > > >

> > > > There is a topology northbound of C and E (not shown) and a topology

> > southbound of A (not shown).

> > > > Cost on all links is 10 except B-D and D-E where cost is high.

> > > >

> > > > C is a node with 1000 neighbors.

> > > > When all links are up, shortest path for all northbound destinations is via

> > C.

> > > > All nodes in the network support fast flooding except for Node D.

> > > > Let’s say fast flooding is 500 LSPs/second and slow flooding (Node D) is 20

> > LSPs/seconds.

> > > > If  Node C fails we have 1000 LSPs to flood.

> > > > All nodes except for D can receive these in 2 seconds (plus internode

> > delay time).

> > > > D can receive LSPs in 50 seconds.

> > > >

> > > > [Bruno] Thanks for your example. Agreed so far.

> > > >

> > > > When A and B and all southbound nodes receive/process the LSP

> > updates they will start sending traffic to Northbound destinations via D.

> > > > But for the better part of 50 seconds, Node D has yet to receive all LSP

> > updates and still believes that shortest path is via B-C. It will loop traffic.

> > > >

> > > > [Bruno] May I remind you that we are discussing IS-IS flooding in order to

> > sync LSDB (LSP database). That is already a big enough subject. It does not

> > including FIB (updates), nor IP forwarding.

> > > >

> > > > Quoting you “when only some nodes in the network support faster

> > flooding the behavior of the whole network may not be "better" when faster

> > flooding is enabled because it prolongs the period of LSDB inconsistency.”

> > > >

> > > > Taking your own examples, in both cases (all nodes support fast flooding;

> > all nodes but D support fast flooding) the period of LSDB inconsistency is 50

> > seconds. Hence this example does not illustrate your statement.

> > > >

> > > > Hence I’m restating my questions:

> > > >

> > > > > > when only some nodes in the network support faster flooding the

> > behavior

> > > > > of the whole network may not be "better" when faster flooding is

> > enabled

> > > > > because it prolongs the period of LSDB inconsistency.

> > > > >

> > > > > 1) Do you have data on this?

> > > > >

> > > > > 2) If not, can you provide an example where increasing the flooding

> > rate on

> > > > > one adjacency prolongs the period of LSDB inconsistency across the

> > > > > network?

> > > >

> > > >

> > > > Had all nodes used slow flooding, it still would have taken 50 seconds to

> > converge, but there would be significantly less looping. There could be a

> > good amount of blackholing, but this is preferable to looping.

> > > > [Bruno] You are using an example where ordering FIB updates across the

> > network, e.g. as per [1], allows to reduce _FIB_ inconsistency across the

> > path/network. And you seem to conclude from this that this translates to

> > LSDB update ordering. Those are two different things. In this thread, I’d

> > suggest that we focus on IGP flooding and LSDB sync only. (*)

> > > > [1] https://tools.ietf.org/html/rfc6976

> > > > (*) We can discuss loop free IGP converge in a different thread if you

> > want. IMO, the use of segment routing/source routing is better than oFIB.

> > But at some point, it still relies on fast flooding when multiple LSPs are

> > involved. (and I mean _fast_ not _ordered_)

> > > >

> > > > --Bruno

> > > >

> > > > One can always come up with examples – based on a specific topology

> > and a specific failure - where things might be better/worse/unchanged in the

> > face of inconsistent flooding speed support.

> > > > But I hope this simple example illustrates the pitfalls.

> > > >

> > > >     Les

> > > >

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

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

> > > > > Sent: Tuesday, May 05, 2020 8:28 AM

> > > > > To: Les Ginsberg (ginsberg) <[e-mail]<mailto:[e-mail]>>; [e-mail]<mailto:[e-mail]>

> > > > > Subject: Flooding across a network

> > > > >

> > > > > Les,

> > > > >

> > > > > > From: Lsr [mailto:[e-mail]] On Behalf Of Les Ginsberg

> > > > > (ginsberg)

> > > > > > Sent: Monday, May 4, 2020 4:39 PM

> > > > > [...]

> > > > > > when only some nodes in the network support faster flooding the

> > behavior

> > > > > of the whole network may not be "better" when faster flooding is

> > enabled

> > > > > because it prolongs the period of LSDB inconsistency.

> > > > >

> > > > > 1) Do you have data on this?

> > > > >

> > > > > 2) If not, can you provide an example where increasing the flooding

> > rate on

> > > > > one adjacency prolongs the period of LSDB inconsistency across the

> > > > > network?

> > > > >

> > > > > 3) In the meantime, let's try the theoretical analysis on a simple

> > scenario

> > > > > where a single LSP needs to be flooded across the network.

> > > > >

> > > > > - Let's call Dij the time needed to flood the LSP from node i to the

> > adjacent

> > > > > node j. Clearly Dij>0.

> > > > > - Let's call k the node originating this LSP at t0=0s

> > > > >

> > > > > >From t0, the LSDB is inconsistent across the network as all nodes but k

> > are

> > > > > missing the LSP and hence only know about the 'old' topology.

> > > > >

> > > > > Let's call  SPT(k) the SPT rooted on k, using Dij as the metric between

> > > > > adjacent nodes i and j. Let's call SP(k,i) the shortest path from k to i; and

> > > > > D(k,i) the shortest distance between k and i.

> > > > >

> > > > > It seems that the time needed:

> > > > > - for node j to learn about the LSP, and get in sync with k, is D(k,j)

> > > >
--
Gyan  Mishra
Network Engineering & Technology
Verizon
Silver Spring, MD 20904
Phone: [phone]
Email: [e-mail]<mailto:[e-mail]>



_________________________________________________________________________________________________________________________

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.