Comments on: Tuning OSPF Performance

By: raghavme

raghavme — Thu, 17 Jun 2010 19:30:32 +0000

Hello Petr,

Is it possible to tune the throttle timers for a particular subnet only?

In my case, I have hub-spoke OSPF topology. Areas 1,2,3,4,5,6,7 are connected to Area 0 via Leased Line.

However, there are 100 VSAT links terminated on the core router in Area 0. These links are very unstable (8kbps,600-1000ms response time)

Is it possible to tweak the timers only for this particular subnet, so that the up/down events are not registered as much.

The bandwidth on these links are also very less, so when the routing updates are exchanged, it takes around 2-5 minutes for OSPF convergence to complete.

By: OSPF Fast Convergence | CCIE Blog

OSPF Fast Convergence | CCIE Blog — Wed, 02 Jun 2010 22:11:02 +0000

[...] Router ID) by the router. You may find more information about truncated exponential backoff in [TUNING-OSPF], but in short the process works as following. Three parameters control the throttling process: [...]

By: OSPF Performance - So Do You Want to be a CCIE? - 2bccie.com

OSPF Performance - So Do You Want to be a CCIE? - 2bccie.com — Sat, 22 May 2010 09:34:46 +0000

[...] According to ine bog: http://blog.ine.com/2009/12/31/tuning-ospf-performance/ timers throttle spf start increment max_wait:is used when dynamic adaptation to unstable network [...]

By: Petr Lapukhov, CCIE #16379

Petr Lapukhov, CCIE #16379 — Sun, 03 Jan 2010 06:08:38 +0000

@Walter

Thanks for the correction, i’ve edited the blog post. When you enable iSPF the amount of memory used would be inremented by a number proportional to 2*N where N is the number of routers in the area. Thus, even for largest areas it should be virtually insignificant for modern routers.

The only potential problem might be some sort of programming bug in iSPF implementation, but the feature has been there for quite some time and appears to be mature now.

With iSPF or without it, SPF computations are not a big issue for modern routers; for large LSDBs and routing tables the main impact might be updating the FIB with new route information.

By: Walter Gibbons

Walter Gibbons — Sat, 02 Jan 2010 06:35:47 +0000

Excellent article Petr! The the discussion of Dijkstra’s algorithm performance brought back nightmares from my Algorithms class and the big O notation from college.

In the version of IOS 12.4 I am running iSPF is off by default. You have to turn it on if you want to use it. Any idea the impact on memory usage that is incurred by turning it on in a med to large OSPF implementation? Is it a no-brainer to just turn it on?

Thanks!

By: Yasir Ashfaque

Yasir Ashfaque — Fri, 01 Jan 2010 13:58:29 +0000

good explanation..

By: Petr Lapukhov, CCIE #16379

Petr Lapukhov, CCIE #16379 — Fri, 01 Jan 2010 07:33:25 +0000

Hey wait a sec, Dijkstra invented SPF, but John Moy designed OSPF! Both did a tremendous amount of work, however I mostly know Dijkstra for his incredible work in formal program verification and the wonderful book “A Discipline of Programming”.

By: amit chopra

amit chopra — Fri, 01 Jan 2010 07:13:37 +0000

ok seriously tell me petr ” Who are you?”

Surely you are the incranation of Mr Edsger Dijkstra who wrote OSPF