Recently, we recorded a webinar to explain a design concept frequently used by iparchitechs.com to build and migrate WISP, FISP and Telco networks – separation of network functions. It centers around simplification of roles within an ISP network. It also explores the use of lower-cost commodity network equipment to maximize the service area for a given ISP footprint while meeting key requirements like scale, redundancy and capacity.
Topics that were covered include:
What are network functions?
Design examples for WISP/FISP and Telco
Equipment and budget considerations
Here is an example of solving design/operational issues with network function separation:
VPLS is a pretty common technology in ISPs to either sell layer 2 services or backhaul traffic to a centralized aggregation point to conserve IPv4 space; check out more on that here.
How can I take advantage of segment routing but still deliver the same services? We already looked at how to setup the label switched paths utilizing IP Infusion as a segment routing mapping server in this interop post. Now let’s see how we can deliver a VPLS service over this with mikrotik as a provider edge router.
Delivering a service with a L2VPN
After setting up the IGP and label distribution between the PEs we will start building the L2VPN.
Why a BGP signaled VPLS session instead of LDP signaled VPLS?
In the segment-routing domain there is no LDP running. However, some vendors support static pseudowires or other methods to bring up a targeted LDP session for VPLS. I did some basic testing here couldn’t easily identify the right combination of knobs to make this work. Don’t worry; I’ll come back to it.
BGP signaled VPLS is a standards based technology that both vendors support.
First thing we need to do after having loopback reachability is to build the BGP sessions.
On OcNOS the route-target and route distinguisher is automatically derived from the vpls instance creation. While you explicitly set it on MikroTik.
ipi-1.lab.jan1.us.ipa.net# show mpls vpls TEST
Virtual Private LAN Service Instance: TEST, ID: 10
SIG-Protocol: BGP
Route-Distinguisher :65000:10
Route-Target :65000:10
VE-ID :10
The site-id (Tik) or VE-ID (OcNOS) are the vpls endpoint identifiers which uniquely identify each PE.
The final step before we can start to test is to setup the attachment circuit. On OcNOS we will have to setup a service template to pair with the vpls instance on attachment circuit. Then you can assign the vpls instance and the service template to a switchport. We will match vlan tag 10 on our service template. Anything coming in with dot1q 10 from the CCR2004 will enter the VPLS tunnel.
ipi-1.lab.jan1.us.ipa.net:
service-template TEST
match outer-vlan 10
!
interface xe2
switchport
mpls-vpls TEST service-template TEST
exit-if-vpls
!
On the MikroTik side we need to setup a bridge for the autocreation of the tunnels and assign appropriate interfaces to it. Again, we will match on dot1q 10.
Finally, we can do testing and verification. We want to look for the mesh peer which is the autodiscovered PE also in this vpls instance.
ipi-1.lab.jan1.us.ipa.net#show mpls vpls TEST
Virtual Private LAN Service Instance: TEST, ID: 10
SIG-Protocol: BGP
Route-Distinguisher :65000:10
Route-Target :65000:10
VE-ID :10
Attachment-Circuit :UP
Learning: Enabled
Control-Word: Disabled
Group ID: 0, Configured MTU: 1500
Description: none
service-tpid: dot1.q
Operating mode: Raw
Configured interfaces:
Interface: xe2
Service-template : TEST
Match criteria : 10
Mesh Peers:
100.127.2.0 (Up)
Then we want to make sure we are learning mac-addresses. As you can see the MACs not learned from the attached device are learned from the BGP peer at 100.127.2.0.
Next we can test reachability. I did have a strange issue where if the AC went down on the MikroTik-1 the mesh wouldn’t form on reattachment. This was resolvable with the following command on IPI-1: clear bgp 100.127.2.0 l2vpn vpls
