Saturday, May 9, 2015

Remote Loop Free Alternate

In the below topology, in case of failure of link between S and D, S will forward the traffic to A and A will again forward the traffic to S. The reason is very simple the cost of next hop from A to D will be lesser than the alternate path which is A-B-C-D-E. This will create a micro loop till the topology will converge.

So answer is very clear, IP FRR is not the viable option in case of ring and square topology. Instead of LFA, Remote LFA is used when no Local LFA is available and is used to find a remote node which can reach the protected neighbor without traversing the protected link. Remote LFA is primarily used for ring and square based access topologies.

In Loop-Free Alternates, the backup routes (repair paths) are pre-computed and installed in the router as the backup for the primary paths. Once a router detects a link or adjacent node failure, it switches to the backup path to avoid traffic loss. Remote LFA allows the backup next-hop to be more than one hop away. It means that after a failure an adjacent node recognizes it and tries to find a (remote) node whose shortest path to the destination is not affected by the failed component. If such node is found then packets will be forwarded to it. Remote LFA relies on tunnels to provide additional logical links towards backup next-hops. After the remote node receives the package it sends it towards to the primary destination. Note that the tunneled traffic is restricted to shortest paths just like “normal” traffic; hence the tunnel must avoid the failure as well.

Now the question arises how to select the node which can be used to terminate the tunnel. B can’t be selected because it is ECMP for D. C can be the best suitable option to terminate the tunnel. Selection of C can be done by using P Space Q.

Click Here To Read Rest Of The Post...

Thursday, May 7, 2015

How to Calculate Fast Re Route(FRR) - Repair Path

The goal of IP Fast Reroute (IP FRR) is to reduce failure reaction time to sub-50ms by using a pre-computed alternate next-hop, in the event that the currently selected primary next-hop fails, so that the alternate can be rapidly used when the failure is detected. The alternate path or repair path must be loop free path to destination in case of network failures without recomputing the network topology as per below mentioned design.

Repair path can be used for diverting live traffic flowing through the primary path protected by the repair path only if the repair path both bypasses local failure (for example, failing link) and does not cause traffic to loop. Such repair paths are called “loop-free alternatives” or LFA.

Repair path provides loop-free alternative if following criterion is met:
1. Loop Free Alternate:- Condition: D(N,D) < D(N,S) + D(S,D). Path is loop-free because N’s best path is not through local router.” Traffic sent to backup next hop is not sent back to S.
2. Downstream Path:- Condition D(N,D) < D(S,D). “Neighbor router is closer to the destination than local router.” Loop-free is guaranteed even with multiple failures (if all repair-paths are downstream path).
3. Node protection:- Condition D(N,D) < D(N,E) + D(E,D). “N's path to D must not go through E.” “The distance from the node N to the prefix via the primary next-hop is strictly greater than the optimum distance from the node N to the prefix.“

It will be more cleared with below stated example:-

Examples of Calculating Fast Re-Route

Click Here To Read Rest Of The Post...

Tuesday, March 10, 2015

Convergence Delay of a Routing Update For Static, RIP, OSPF, EIGRP, ISIS and BGP

Got some good docs for the maximum Convergence Delay of a Routing Update For Static, RIP, OSPF, EIGRP, ISIS and BGP.
Br/> Using the below described convergence points the theoretical maximum propagation delay of a routing update can be calculated from one VPN site to another VPN site (CE router to CE router).

- The convergence points are taken as T1_max, T2_max, T3_max, T4_max till T8_Max

- As per the convergence points the maximum theoretical convergence times are calculated for various protocols like RIPv2, EIGRP, OSPF, ISIS, BGP and static routes.

- The total convergence time taken for a static route is 25 seconds

- The maximum convergence time for RIP approximately 85 seconds for new routes, 25 seconds for serial links and 265 seconds in case of route flap.

- For EIGRP the convergence timer is much faster and is around 25 seconds.

- The convergence time increases slightly to 35 seconds in case of OSPF

- For ISIS the maximum theoretical convergence value is same as OSPF approximately 35 seconds.

- The convergence value increases to 85 seconds in case of BGP.

Click Here To Read Rest Of The Post...