This post is a part of my preparation to JNCIP-SP exam, which covers a lot of service provider stuff.
Due that preparation a virtual machine has been prepared, based on Unetlab (UNL).
What is UNL and how to cook it may be discovered on official site or that (RU) comprehensive article.
VM consists of 5 Juniper MX routers (v14.1) as SP devices and 3 Cisco 3725 routers (advipk9-12.4) as CEs.
This is how it looks like in UNL web interface:
Beyond you may find all further infomation about what is included and how to use it.
UPDATE: new version of the lab is available here – http://sk1f3r.ru/jlab. Memory requirement has been decreased to 4GB (to be honest even less: 256MB+2G+(256MB*3)=3G).
Topology depicted in Visio as well (click to enlarge) to provide more information and allow to evolve the network as a scheme at first.
Source Visio file link – here (not available anymore, please, use the last version of the lab – http://sk1f3r.ru/jlab).
Download link for the VM – UNETLAB_Juniper-SP_160106_00.ova (not available anymore, please, use the last version of the lab – http://sk1f3r.ru/jlab).
- Virtualization software: VMware Player/Workstation/ESXi, Oracle VirtualBox;
- VM Memory amount: 16GB recommended (10GB minimum);
- VM NIC amount: 1 (bridged);
- VM disk drive size: 20 GB.
16 GB DRAM is the requirement because basically an MX router consumes 2GB of memory and Cisco 256 MB = 10GB+
The file provided above is an OVA file and that means you can import the VM in your virtualization software without magic moves.
Double click on the file, type machine name, wait few moments till coping is done and we are ready to start.
Start VM and wait till login prompt:
In case you need access to UNL CLI use such credentials:
As system says, use provided link to get access to the web interface.
Credentials for the web interface (default):
Press Labs in top menu, then LABS below in Folders section and MPLS-LAB.unl in Labs section.
In top menu select Actions and start all nodes as shown below.
All routers were powered on and by clicking on these icons we can get access to a device console.
The topology depicts a service provider with 5 devices. 3 of them are PE devices, which means they are facing with customers devices and rest routers are exists for SP infrastracture itself. The following names and roles assigned to SP devices:
- PE1 – facing Site 1 of Customer 1;
- PE2 – facing Site 2 of Customer 1;
- PE3 – facing Site 2 and 3 of Customer 1;
- P – core device for interconnect purposes;
- P-RR – core device with BGP RR role.
Credentials for console access are the same as for UNL CLI.
The following services provided to Customer 1 by SP:
-  L2VPN-M – MPLS L2VPN Martini – RFC4906;
-  L2VPN-K – MPLS L2VPN Kompella – RFC6624;
-  VPLS-B – VPLS BGP signaling – RFC4761;
-  VPLS-L – VPLS LDP signaling – RFC4762;
-  L3VPN – MPLS L3VPN BGP – RFC4364.
All such serviced are depicted in the Visio scheme exactly as it names in the list above and uses unit number, VLANID, instance identifiers equaly to service ID as shown in  brackets.
SP IP connectivity information listed below.
The following technologies are in use:
- BGP Route Reflection;
- RSVP Automatic mesh;
- Fast Reroute (FRR);
- MPLS Automatic Bandwidth Allocation;
- VPLS Multi-homing (for VPLS-B);
- BGP as the Provider/Customer Edge Protocol for BGP/MPLS IP Virtual Private Networks;
- OSPF as the Provider/Customer Edge Protocol for BGP/MPLS IP Virtual Private Networks.
5. What to do next
The provided topology is useful for engineers which looking for an environment to get some hands-on expirience with common MPLS services based on Juniper equipment. Articles across the Internet mostly covers MPLS services as separated things, but this topology may be used to explore them all in a single network. An engineer may dive in services without spending hours for an interconnect configuration as well.
- explore how SP infrastructure works (LSPs, RIBs, FIB, TED, etc);
- test an network reliability (FRR, VPLS Multi-homing switchover, etc);
- delete services and configure back;
- switch from dynamic LSPs to manual;
- switch from OSPF to IS-IS;
- add more dynamic protocols to CEs.
VM eventually created and exported with VMware Workstation software. I’ve encountered with virtualization problems on last Oracle Virtualbox release, when software says that virtualization is enabled but UNL VM cant recognnize it.
Sometimes, core devices (P/P-RR) console stops to react on buttons, but the topology continues to work as intended. You may manually stop/start the device to get access back.
It is recommended to disable PE-interfaces to test reliability for things like multi-homing, due CE devices do not support BFD (IOS 12.4) and UNL connects interfaces not directly, so a disabling of interfaces on one side do not cause link down on other.
This section provide the infomation about plans to change the topology with an explicit notice if it is done.
Next few changes:
- enable IS-IS across SP network. OSPF will still be in use with lower route preference, but it is easy to deactivate ospf stanza if want to use IS-IS as a main IGP;
- add p2mp for VPLS;
- add EVPN service with ID 401 – RFC7432;
- your recommendations are welcome.
[!] My excuses for such a poor language. I’ll improve some explanations in future to make the article more viable for my english and russians readers. TY
[!] Anyone may use the text and pictures in this article without mentioning the author and reference to the source.