On Mon, 2 Sep 2002, Petri Helenius wrote:
"Stephen J. Wilcox" wrote:
On Mon, 2 Sep 2002, Petri Helenius wrote:
bdragon@gweep.net wrote:
With link-state, one interface flap can mean doing SPF on every route.
Only if you learned every one of your routes from different neighbor. If you have two exits and 100000 routes, you calculate twice and apply the results to the prefixes.
Note that this does not apply to a proprietary, "hybrid", semi-link state protocol marketed with name "EIGRP" where all routes need per-prefix calculation. (OSPF and IS-IS work fine)
but.. with SPF you need to run the algorithm on all paths for each flap and then see what that does to your routes with eigrp you only need to apply the algorithm to any route on the link that flapped and then only on the attached router (which will propogate much like bgp if it requires other routers to recalculate) yes thats bad if yuo have 100000 routes but you shouldnt have! assuming a smaller routing table yuo get quicker convergence and much much less CPU requirement on your rotuers Steve
Pete
Steve
Pete
"Stephen J. Wilcox" wrote:
but.. with SPF you need to run the algorithm on all paths for each flap and then see what that does to your routes
Only the paths that cross the one you lost. Obviously if this happens or not, depends on your implementation. Look in the documentation under heading "partial SPF".
with eigrp you only need to apply the algorithm to any route on the link that flapped and then only on the attached router (which will propogate much like bgp if it requires other routers to recalculate)
For this you need to look under the heading "non-technical marketing propaganda". So far, I've yet to figure out a real life network example where EIGRP would outperform either OSPF or IS-IS. Obviously if you compare an ancient OSPF implementation to a more recent implementation of another protocol, differences in performance can be found. Objective measurements are hard to come by.
yes thats bad if yuo have 100000 routes but you shouldnt have! assuming a smaller routing table yuo get quicker convergence and much much less CPU requirement on your rotuers
Most networks have more routes than they have links or routers. This already by the nature that most links have one route. So the assumption that number of routes would be small is usually incorrect. Obviously if you're talking about <50 routes, choice of routing protocol does not make a difference and you'd be probably happy running RIPv2. Pete
With link-state, one interface flap can mean doing SPF on every route.
Only if you learned every one of your routes from different neighbor. If you have two exits and 100000 routes, you calculate twice and apply the results to the prefixes.
Note that this does not apply to a proprietary, "hybrid", semi-link state protocol marketed with name "EIGRP" where all routes need per-prefix calculation. (OSPF and IS-IS work fine)
but.. with SPF you need to run the algorithm on all paths for each flap and then see what that does to your routes
with eigrp you only need to apply the algorithm to any route on the link that flapped and then only on the attached router (which will propogate much like bgp if it requires other routers to recalculate)
yes thats bad if yuo have 100000 routes but you shouldnt have! assuming a smaller routing table yuo get quicker convergence and much much less CPU requirement on your rotuers
And with nailed BGP routes you dont need additional layer of complexity. Alex
participants (3)
-
alex＠yuriev.com -
Petri Helenius -
Stephen J. Wilcox