Nokia Lab | LAB 8 RSVP-TE Resiliency |


Hello!

Today I'm going to lab one of my favorite topic - RSVP-TE Resiliency. It's a real pleasure to see how different methods of network resiliency can make your network more stable and reliable. 

Please check my first lab for input information.

Topology example

Lab tasks and questions:
  • Preparing
    • create LSP to_R6 on R1 (parameters: "totally loose" path, cspf)
    • create LSP to_R1 on R6 (parameters: "totally loose" path, cspf)
    • create Epipe service between CPE1 and CPE6. It helps to compare convergency time in the different cases. I use Virtual PC nodes as CPE devices and simple ping as a tool. It's not a production-ready tool suite, but it's relevant for the education lab. (Depend on your CPE devices, but I recommend use rapid ping or adjust send/receive timers. Results will be more clearly)
    • configure BFD (I use TX/RX equal 100ms):
      • on L3 interfaces
      • on OSPF interfaces
      • on RSVP interfaces
    • configure IP addresses on CPE
    • check IP connectivity between CPE
  • Secondary paths
    • The first method is Non-standby secondary path. Add secondary path to exist LSPs
      • check secondary path operation status
      • our goal is investigation the reconvergence process
        • run ping on CPE and break the primary path of any LSP(you can shut down ports, interfaces, intermediate router, etc)
          • examine LSP and paths status (You can look at LSP to_R6 or to_R1. It doesn't matter)
          • describe the reconvergence process
    • The second one is Standby secondary path. Add option standby to the secondary path. So let's do the same steps to investigate the reconvergence process in this case
      • check actual hops for primary and secondary paths
      • run ping on CPE and break the primary path of any LSP
        • I'm sure that the results are unsatisfactory. Very likely primary and secondary path have the same list of actual hops. And probably we will get a full outage duration of 30 sec - retry timer. As a conclusion of this test - if we use standby secondary path, we must distinct primary and secondary path. And we can use the next methods: explicit path, admin-groups, SRLG. Let's discover SRLG.
      • SRLG
        • Our goal is the path diversity by SRLG. The first step is the creation of SRLG groups (e.g. "RED" and "GREEN"). Then we should mark some interfaces. Everything depends on network topology. Previously I decided to mark interfaces R1-R2, R2-R4, R4-R6 in "RED" and mark interfaces R1-R3, R3-R5, R5-R6 in "GREEN". But I may get a situation when the primary path goes through "RED" and "GREEN" interfaces simultaneously (e.g. R1 - R3 - R4 - R6). And as a result, Head-end can't signal the secondary path. We will get the error - noCSPFroute. I decided to increase IGP metric between R3 and R4. It helps to distinguish paths.
          • examine path diversity by SRLG  and reconvergence process in this case
    • Switchover between secondary paths
      • create strict path sec_strict with explicit hops - R1 - R3 - R4 - R6 on R1
      • create strict path sec_strict with explicit hops - R6 - R4 - R3 - R1 on R6
      • bind new secondary path to LSPs - to_R1 and to_R6 and enable the standby option 
        • investigate reconvergence process in this case
          • examine the secondary path selection criteria (Try to change path priority, standby option, configuration order)
  • Fast Reroute
    • One-to-One Backup Method
      • shutdown LSPs to_R1 and to_R6 and create new ones with Fast-reroute one-to-one option(e.g. to_R6_FRR_onetoone, to_R1_FRR_onetoone)
        • examine detour signaling process
          • What is the object Head-end includes in the Path message for detour signaling?
        • examine reconvergence process
          • What is the main difference between Fast Reroute and Secondary path protection method? What method is faster if we speak about network repair? Why?
          • add Secondary path("totally loose") to LSP and repeat investigation
            • describe reconvergence steps in this case
            • What benefit will we get with the combination FRR and Secondary path?
    • Facility Backup Method
      • shutdown LSPs to_R6_FRR_onetoone and to_R1_FRR_onetoone and create new ones with Fast Reroute Facility option
        • examine Bypass tunnels signaling process
          • describe differences between Detour and Bypass tunnels 
          • describe the meaning of PLR and MP
          • create a new LSP from R1 to R6 (or vise versa) and examine how Bypass tunnels protect several LSP
      Final configs
      Github project

      Useful debug commands
      show router rsvp session detail
      show router mpls lsp detail
      show router mpls path detail
      tools dump router mpls te-lspinfo detail
      show router bfd session

      Comments

      Post a Comment