Header Only - DO NOT REMOVE - Extreme Networks
Solved

Customer wants Q-in-Q VLAN Transparency

  • 27 June 2019
  • 1 reply
  • 407 views

I need some assistance with Q-in-Q design. I have a full Layer 3 network running OSPF.

How can I offer VLAN transparency for my cuistomers?

Would Ethernet over MPLS work and what would the configuration look like.

I am running MLXe in my environment. IronWare : Version 6.3.0aT163
icon

Best answer by Steve Worrall 28 June 2019, 16:59

Hello,

I assume that you want to be able to carry your customer VLANs as a virtual Ethernet service - like a pseudowire.

In that case the MLX can support that. The configuration can be quite simple especially if you use LDP as the signalling for the MPLS labels.

You should look in the MPLS configuration guide for full details, particularly the VLL section (virtual leased line).

Broadly the MPLS config will look something like on your edge routers. Transit routers would just have the mpls-interfaces defined. These can be physical ethernets or ve interfaces.

router mpls
mpls-interface ethernet 0/1
ldp-enable
mpls-interface ethernet 0/2
ldp-enable

vll my_vll 1001
peer 10.0.0.2
untagged ethernet 0/3


The peer IP address that you see in the vll configuration will need to be the lo1 IP address on the other router. lo1 will need to be advertised within OSPF so that it is reachable.

You will also need to enable MPLS on the interfaces of all routers in your network that are likely to be transit routers for the VLL - usually all routers.

Ethernet 0/3 is the port where you connect your customer interface. To enable it to pass all VLANs you will need to change the VLAN ethertype that it is expecting to see. We often set this to 9100 as it is commonly not used as an Ethertype.

tag-type 9100 eth 0/3

MPLS configuration can become very involved depending on what you want to achieve - this would usually result in you using RSVP for signalling, in which case you can start to define the links that are used and also fast failover. Using LDP as a start will get you going quite simply. Failover of the VLL to a new link (in case of failure) would probably take a second or two (compared to using RSVP and LSPs where the failover could be less than a tenth of a second)

Steve
View original

1 reply

Userlevel 1
Hello,

I assume that you want to be able to carry your customer VLANs as a virtual Ethernet service - like a pseudowire.

In that case the MLX can support that. The configuration can be quite simple especially if you use LDP as the signalling for the MPLS labels.

You should look in the MPLS configuration guide for full details, particularly the VLL section (virtual leased line).

Broadly the MPLS config will look something like on your edge routers. Transit routers would just have the mpls-interfaces defined. These can be physical ethernets or ve interfaces.

router mpls
mpls-interface ethernet 0/1
ldp-enable
mpls-interface ethernet 0/2
ldp-enable

vll my_vll 1001
peer 10.0.0.2
untagged ethernet 0/3


The peer IP address that you see in the vll configuration will need to be the lo1 IP address on the other router. lo1 will need to be advertised within OSPF so that it is reachable.

You will also need to enable MPLS on the interfaces of all routers in your network that are likely to be transit routers for the VLL - usually all routers.

Ethernet 0/3 is the port where you connect your customer interface. To enable it to pass all VLANs you will need to change the VLAN ethertype that it is expecting to see. We often set this to 9100 as it is commonly not used as an Ethertype.

tag-type 9100 eth 0/3

MPLS configuration can become very involved depending on what you want to achieve - this would usually result in you using RSVP for signalling, in which case you can start to define the links that are used and also fast failover. Using LDP as a start will get you going quite simply. Failover of the VLL to a new link (in case of failure) would probably take a second or two (compared to using RSVP and LSPs where the failover could be less than a tenth of a second)

Steve

Reply