Redistribution

silver145

Hi all,

Redistribution seems to be one of those annoying ones for me currently, just wondering what you have done to slap this topic in the chops?

lrb

Routing TCP Volume 1, some vendor's workbooks (INE, Narbik, IP Expert), and whole lot of labbing.

Learn the caveats of each protocol handles the metrics of redistributed routes (OSPF type 1, type 2, default cost; EIGRP default metric, per-prefix vector metrics; RIP default metric, per-prefix metric; BGP MED, community encoding)

Learn route maps like the back of your hand. Learn how to use sensible route tag numbers, where to apply them and where to block them.

Learn how routing loops can occur when doing mutual redistribution at multiple points in the network.

Make lots of mistakes Nothing motivates you to learn how to redistribute properly after you have caused an outage due to a loop!

And if worst comes to worst in the lab and you are losing too much time, simply perform mutual redistribution at only a single point in the network so end-to-end connectivity can still be achieved.

gorebrush

Honestly? Very little yet.

I'm more concerned about nailing the protocols as units first but obviously this is an area I need to revisit. I know how redistribution works though so will just be something I concentrate on labbing wise, i.e. I'll practice it a lot

silver145

Cheers for the responses you two,

I am fairly confident with most of the initial processes, defaults etc. But when i build my own lab i am building it already knowing where loops will form and how so i'm looking for any resource that suggests topologies and basically says "find the problem" as this way i wont have prior knowledge of where the loop is going to be and it gives me "on the spot" evaluation time

gorebrush

Well I would consider the topology and be looking for multiple entry and exit points between the different routing protocols - when there is a single point then there is no danger of creating a loop or getting route feedback.