We currently have an MPLS-VPN network in our internal network for couple of clients. We have fully redundant links between PE and P routers. Our internal routing protocol is EIGRP and MP-BGP for PE-to-PE peering. However, let's say for example PE1 to PE2 is passing through PathA. It chose this path due to internal routing metrics but there's an alternate path which is PathB which has uglier metric. If there's a failure in PathA, the alternate path will be used. However, when all the network links along PathA are back, it is causing downtime to the clients connected between PE1 and PE2 VPNs. It takes more than 30 seconds or even a minute for the service to go back. I'm trying to check the MPLS forwarding table and RIB during the downtime but I cannot figure out what's causing the delay. Is there a way for MPLS to make sure it would wait for something and make sure PathA would be a good path before traffic passes through?
there are many solutions to this kind of issues like Fast Reroute , MPLS TE to balance the traffic or better administer the metrics and so on. The point anyway is that the 30 seconds of delay are most likely due to the IGP reconvergence time used by default by mpls to build the LFIB. I strongly suggest you to implement BFD on each link of your P-Network up to the PEs and to start using MPLS TE for FRR (Fast Reroute) . This would avoid the all the traffic to PathB to switch back to PathA at once.
Being the P-Network the most important part of an SP in terms of service i would even suggest you to move to OSPF or IS-IS like your P-Network routing protocol. Prioritize BFD by now !
It appears that when the links along the PathA came up EIGRP adjacencies where formed almost instantly and traffic was switched to PathA links.
However LDP adjacencies where not formed yet or haven’t got enough time to exchange labels.
So until the LDP adjacencies are formed, labels exchanged and CEF programed with labels traffic on PathA is being dropped.
You can avoid this using LDP IGP synchronization
With LDP IGP synchronization if IGP adjacency is formed on a particular link where LDP is enabled it will wait for LDP to form it’s adjacency or IGP will wait for some configured "wait time" – during this "wait time" it will advertise the link prefix as well as all prefixes learned over this link with maximum metric so that this link is never used as best path – so the "wait time" should be enough for LDP adjacencies to be formed, labels exchanged and CEF programed with labels, so once the IGP starts to advertise the link and all prefixes learned via this link with "normal" metric and there’s traffic forwarded via this new path it can be label switched successfully.
However I’m afraid that LDP IGP synchronization is available only for ISIS or OSPF
This document is an early notification of a behaviour change that will be introduced in IOS XR release 6.5.
IOS XR configuration principles relevant for this article are:
On router platforms all interfaces must be by defaul...
With XR 4.2.0 the ASR9000 is releasing a new line of hardware models. This amongst others is the RSP440, the next generation RSP with faster switch fabric along with Typhoon based Linecards, the next generation network processor.
The Cisco EPN system incorporates a network architecture designed to consolidate multiples services on a single Multiprotocol Label Switching (MPLS) transport network. This network is designed primarily based on Application ...