Netonix Forums

Anyone using edgemax at pops? I'm considering them, I have a few that are not used tower side, but at customers or at my CO for VPNs etc. Currently all 'tik for routers.

I feel like 'tik is falling behind on this. Their CCR series is very rack oriented and I don't think suitable for tower-side. My workhorse, the rb493g is just not going to be up to the task of 'ac based pops.

The ER series have the pps and throughput to handle the next-gen stuff...


just figured I'd start a thread to get this section going :)

The edgeMAX routers are pretty nice.

We tried to use them but had to pull them because we use OSPF with BFD and they do not yet support BFD, hopefully soon but the Vyatta branch they used does not support BFD and they would have to port all the code. I heard they were considering it?

BFD because you weren't satisfied with OSPF convergence times I assume?

rebelwireless wrote:BFD because you weren't satisfied with OSPF convergence times I assume?

Yes, with BFD the convergence time is usually less than 1 ping, even most VOIP calls do not notice the route change.

well, don't hold your breath on ubiquiti adding that this decade :/ Might be easier to do it yourself w/ bfdd. Not sure how modified quagga is in vyatta or how old the version is. In 5 minutes on google I found a lot of resources for that so I don't think it would be terribly hard to implement.

That's why you run true ring topology, not OSPF

mhoppes wrote:That's why you run true ring topology, not OSPF

Whats wrong with OSPF with BFD?

What do you run Matt?

yeah Matt, elaborate. any routing or bridging protocol is going to have convergence times, what is faster about a ring topology (of which you didn't mention which tech) vs OSPF w/ BFD? or MPLS-TE w/ FRR or whatever?

I keep having this mad thought about batman-adv to get full layer2, arp caching with <50ms convergence, multipathing, and simplified topology. :/

I'm just saying OSPF isn't how the "big dogs" do ring fail over.

They run AdTran, Performant (now Acedian), or similar equipment that does <50ms switch hit convergence time between one ring direction and the other.

T-Mobile is running Acedian. We run AdTrans in our network.

Well, I am a puppy I guess but OSPF with BFD gives me the same convergence timing (most times).

There are other protocols and we do plan to explore them someday but for now OSPF with BFD does the job.

OSPFv3 is a whole lot better but it is complex and it is easier if your running IPv6.