cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
796
Views
0
Helpful
4
Replies

L3-MPLS VPN Convergence

shehzadhassan
Level 1
Level 1

Perhaps someone on this group can identify the missing timers/processing-delays in end-to-end client route convergence

Scenarios:

a) BGP New route Advertised by Cleint(CPE1)

b) BGP Route withdrawn by Client(CPE1)

PE-to-RR i-M-BGP (Logical)

========= ----RR------ ======

" | | "

CPE1---->PE1------->P1-------->P2---->PE2----->CPE2

| |

--------->P3-------->P4-------

Routing:

- eBGP btw CPE and PE (any routing prot within Cust site),

- OSPF, LDP in Core,

Timers/Steps I'm aware of:

- Advertisement of routes from CE to PE and placement into VRF

- Propagation of routes across the MPLS VPN backbone

- Import process of these routes into relevant VRFs

- Advertisement of VRF routes to attached VPN sites

- BGP advertisement-interval: Default = 5 seconds for iBGP, 30 for eBGP

- BGP Import Process: Default = 15 seconds

- BGP Scanner Process Default = 60 seconds

Would appreciate if you someone can identify any missing process-delay, timers? specially w.r.t RR.

Thanks

SH

4 Replies 4

amritpatek
Level 6
Level 6

I am just wondering if the scan time on RR would contribute to any delay before reflecting the route.

romccallum
Level 4
Level 4

Change your scan-time and import scan-time to 5 seconds.

fpmw
Level 1
Level 1

Check the LDP/TDP timers in the core. Remember if a link fails in the core, reroute occurs, LDP/TDP binding needs to be renewed. tags are binded on those routes being in the routing table (IGP). So, there is a delay possible from a core prespective:

mpls ldp holdtime

mpls ldp discovery hello [holdtime | interval]

In case you are using TE check these:

mpls traffic-eng topology holddown

mpls traffic-eng signalling forwarding sync

mpls traffic-eng fast-reroute timers promotion

I believe the latter one onyl applies to SDH. In which you use segment loss feature.

Regards,

Frank