logo
painting

(network diagram of the lab)

Juniper vSRX MPLS Lab

I’ve wrote up a lab (available on github) on deploying a MPLS core network using Juniper vSRX routers. It’s completely based on this blog post, entitled “Simple Juniper MPLS Core.”

My goal with replicating the simple MPLS network was to try to get a better understanding of how MPLS will work in NFV and 5G where there is a desire to do “network slicing.”

I’m not sure who wrote the initial blog post, but I am thankful they did because I was able to completely replicate it. I’m indebted to them because I feel like I now have a better understanding of Juniper and MPLS networks, which was my goal. This lab provides a base that I can build off of.

I don’t want to say much more in this blog post because I put a lot of work into the lab, so please take a look at the github repo where it resides.

Here’s a listing of all the virtual routers running an a baremetal KVM node.

$ virsh list
 Id    Name                           State
----------------------------------------------------
 86    ce-ny                          running
 120   pe-lo4                         running
 122   p2-njh                         running
 123   p1-ny8                         running
 124   pe-nj2                         running
 125   ce-uk                          running

Here’s a session of ce-ny pinging/tracerouting ce-uk.

$ ssh root@ce-ny
Password:
--- JUNOS 17.3R1.10 built 2017-08-23 06:47:03 UTC
root@ce-ny%
root@ce-ny% cli
root@ce-ny> ping 6.6.6.6
PING 6.6.6.6 (6.6.6.6): 56 data bytes
64 bytes from 6.6.6.6: icmp_seq=0 ttl=61 time=6.536 ms
64 bytes from 6.6.6.6: icmp_seq=1 ttl=61 time=3.840 ms
^C
--- 6.6.6.6 ping statistics ---
2 packets transmitted, 2 packets received, 0% packet loss
round-trip min/avg/max/stddev = 3.840/5.188/6.536/1.348 ms

root@ce-ny> traceroute 6.6.6.6
traceroute to 6.6.6.6 (6.6.6.6), 30 hops max, 40 byte packets
 1  10.3.5.0 (10.3.5.0)  4.759 ms  4.345 ms  4.514 ms
 2  10.2.3.0 (10.2.3.0)  6.033 ms  6.077 ms  5.291 ms
     MPLS Label=299776 CoS=0 TTL=1 S=0
     MPLS Label=299776 CoS=0 TTL=1 S=1
 3  10.2.4.1 (10.2.4.1)  6.693 ms  6.012 ms  8.049 ms
     MPLS Label=299776 CoS=0 TTL=1 S=1
 4  6.6.6.6 (6.6.6.6)  4.826 ms  5.409 ms  5.222 ms

root@ce-ny>

It works! :)

NOTE: On the diagram each router has a .17x IP address (full address is the management network, 192.168.122.17x). The x for each is also their router ID and loopback address. So ce-uk, .176, is 6.6.6.6.

Conclusion

I’m really happy with how this worked out, and as mentioned, I now have a base to do more exploration around MPLS, BGP, OSPF and other protocols, hopefully ending with me having a good understanding of what network slicing could or should look like in a service provider.

One of the major things I learned is that building a virtual lab like this is really about understanding nodes and edges, just like what one would do with graphviz dot diagrams. This is why I’ve listed the “links” (aka edges) and interfaces on the diagram, to help me automate and validate. Building a network diagram like the above should be completely automatable, based on well understood graphing. But that is another story… :)