I work for a small ISP that’s actually growing some fast. Right now most of the backhaul and core network runs overs a few Cisco routers and switches (asr 9001, some nexus) and a lot of Mikrotiks.
We own a small optical fiber network plus and a Fixed Wireless Access network where most of ours old customers came from. But we also use some external carriers in wholesale mode.
We’re starting to develop a new network model aimed to increase the network overall capacity, Cumulus Linux seems a very good compromise between features and costs.
Here’s the problem:
A lot of our customers are connected to us thru a carrier that delivers the traffic over a QinQ based port.
The service tag defines the pop (may be 5/6 in every big city and one every few small villages) and the COS type (0, 1, 5) so we have, right now, about 400 service tags but in a couple of years this number will be, at least, doubled.
The customer tag defines… The customer.
In theory we can have about 4000 customer under each svlan
Every client/customer’s cpe creates a pppoe tunnel.
In this very moment we create a pppoe server (using a mikrotik ccr and/or chr) for every single customer (carrierPort->SVLAN->CVLAN->PPPoEServerInterface) but our goal is to untag (or strip) all customers vlan and bridge the service vlan into 3 cos-based interfaces/vlan/vxlan tunnels with a single pppoe server for cos 0 (and maybe cos 5)
We got another carrier that delivers traffic from our customer to us using a simple 802.1q trunk port with a tag for every pop and no customer vlan, there are, more or less, 1000 pops in the country. In this case we create a pppoe server for every vlan.
As far as I know, in this case, it should be quite simple to create a vlan-awared bridge between all the vlans in the carrier interface and an interface (or tunnel) to our pppoe server.
Long story short, here’s the question.
Can I do what I need to do using cumulus linux or that’s out of the possibilites?
I know that some Cisco and Juniper equipment accomplish what I’ve descrived, but are both out of our budget at the moment…
Thanks a lot.