Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email firstname.lastname@example.org
From: Mike Yeung (myeungINNOVANCE.COM)
Date: Fri May 18 2001 - 15:17:45 CDT
I have been doing some research in the convergence time within an OSPF
network. It surprise me to see some of the typical number out there being
in the order of minutes (e.g. 2 min) while my own calculations gives me
something in the order of seconds. The following are the setup/parameters
that I used in my calculation and some of the conclusions I came about.
- processor is not busy and routing is running at highest priority
- A partically meshed network with a total of 250 routers.
- Single OSPF area
- Assume reasonable amount of bandwidth for routing operations (e.g. T3)
- Hello interval at 3 seconds
- Router Dead interval at 12 seconds
- No SPF delay of 5 seconds as indicated in Cisco product
- 200 MHz processor
- network converged under 15 seconds
Does anyone think the convergence time I calculated is completely off?
I came to realize the following configurable parameters can greatly
influence the converge time in an OSPF network. Please let me know if you
know of others.
- Hello interval and RouterDeadInterval
- SPF calculation delay (in some Cisco product and Zerba OSPF stack)
- MinLSInterval (only matters if you are dealing with multiple failure at
the same time or unstable failure)