Routing loops

[samcomlink]

may i know how the routing loops problems come in network?

[vmiller]

Generally, when one redistributes between systems is the first case.

[adamboyce]

Loops are very common to distance vector routing protocols.

Fortunately distance vector routing protocols (rip, igrp) implement strategies to avoid routing loops.

Some strategies consist of hold down timers; poison reverse, split horizon, etc

However with all of these techniques human error can still prevail. as the previous post said the most common routing loop situations is when two way redistribution is implemented. Careful consideration is needed if two-way redistribution is used because of the risks it poses (routing loops). The correct way is to do one-way redistribution and default route out.

Regards

[samcomlink]

pls let me know how u remove routing loops means number of ways u can remove loops.ie any command.(OSPF,BGP environment )

[adamboyce]

well it is just carefull planning.

OSPF is quite good when it comes to routing loops.

it quite quickly converges (all routers have the same info) compared to distance vector protocols and loops are quickly identified. however loops generally occour in redistribution The safe way to avoid loops in redistributionof is to only redistribute one way.Your network should be designed heiraticly so if any redistribution occours only redistribute one way(from lowest network up) and default route out for the lower network.

hopes this helps

[ruwhite]

In general, redistribution, network melts, and bad configuration are the causes of permanent routing loops. Temporary routing loops happen in almost all networks.

I think it's something of a myth that routing loops persist longer in a distance vector protocol than a link state protocol--from experience, it seems to me temporary loops will persist in both about the same amount of time.

:-)

Russ.W

[adamboyce]

In a distance vector environment no loops will occur (using Cisco deployments of rip/igrp) due to hold downs and split horizon. The only time networks will not be reachable (that exists) will be when the networks are converging due to a new link or a link going down. That’s where the technique of flash updates and route poisoning are used to speed up convergence

However in Link-state protocols, loops are calculated locally on the router & eliminated (SPF alga, SPF Tree).

Depending on the size and topology the results can vary for time to convergence from protocol to protocol. EIGRP I find the most efficient protocol to converge because of the use of the DUEL algorithm that isolate the update to that specific area of the network (only routers that need to change routes will be affected)

P.S.

Good discussion

[adamboyce]

Sorry should be DUAL

Diffusing Update Algorithm

[samsam_wang]

http://www.cisco.com/en/US/tech/tk365/tk207/technologies_tech_note09186a008009487e.shtml

R1 s0-------s0 R2

R1

ip route 192.168.1.0 255.255.255.0 se 0

R2

ip route 192.168.1.0 255.255.255.0 se 0

now you got a loop

[ruwhite]

How is this redistribution? You can do this with no routing protocols at all....

:-)

Russ.W