basic ospf question!

If you have only two OSPF routers on a broadcast media connected to back to back then you can configure the interface with the osp network type point-to-pont to eliminate the DR election entirely. Dointg so also would the 40 second wait time associated with DR election.

not a lot - DR / BDR is about load management on the CPU, or for handling limited "visibility topologies, such as across a partial mesh ATM network.

Also DR / BDR elections are "sticky" so once you have a DR, it doesnt change unless the current DR goes away, in which case OSPF promotes the current BDR if there is one.

ospf summaries only affect internal routes, and only happen at an ABR as routes cross an area boundary. (OK - a stub area ABR router can aggregate extermals as well, but that is a special case)

the ospf algorithm has explicit limits about routes / LSAs being consistent on all routers inside an area, so there are limits on what can be filtered and where.

you can still use redist to aggregate routes before they get imported into OSPF.

A router summarization is not always possible (for example when you have non-continuous block of subnets), I guess the only option then is to redistribute static/connected or use a route-map to filter stuff.

I was just wondering if we DO have an option to do redistribution or summarization, which one would be faster.

no big deal because we > > If I connect two routers with OSPF in area 0 with the following > > statements

if you just have a few 1000 routes (ie enterprise rather than an internet size routing table) - i doubt it makes any practical difference.

what does have an impact is how many LSAs get propagated after a big topology change - and the easiest way to cut that down is to have fewer routes in the routing table.

in which case summaries of any type are irrelevant unless you want them for other reasons - 1 would be to limit how far a topology change propagates.