Inter-PBX Calling / Best HA Method

Firstly, about me. I've been hosting a single FusionPBX for a couple customers for 4 years now. First it was in a colo on our own hardware, then i moved it to Digitalocean as it was more cost effective for my size. I've read as much as i can here in the forums, and recently signed up as a paid member and starting going through all the info there, as there is hours upon hours of video to sort through, which is awesome.

We have recently merged with another company and are expecting to gain quite a few more customers. Hosting the PBX is as much a business venture as it is a technical experiment for me, so i enjoy figuring out the best or expected way to do things.

This leads me to a few questions, that although aren't important for me now, could be in the future, and i'm curious how everyone else has it laid out.

Firstly is inter-pbx calling, which in turn is an active-active HA setup question i guess. Im in Canada, so i have our system in Toronto. I would like to create a second POP out west, as if a customer has a remote office, or a remote worker, their latency to the Toronto POP gets pretty high, so connecting them somewhere closer would be ideal. I know one of the threads in here, someone was covering an active-active setup, but it stalled it. I wanted to see if anyone else was doing anything to address latency over distance, or is everyone just running active-passive clusters?

Secondly, there are so many ways to skin this cat for HA, its hard to know which is best, and i know that changes for everyone's situation as well. What method is everyone using?
Dual Registration?
DNS SRV?
Smart DNS?
Floating IP?
A combination?

Also, is anyone using any managed cloud options for this now? I'm thinking of moving to Azure, so im wondering if people have used their managed database for the shared backend, or their shared storage for the file sync, or their load balancing for failover? Same applies for AWS and GCP.

My HA/DR is pretty basic right now with a single server obviously, but with the small number of endpoints i have, my RTO of ~5 minutes by restoring a snapshot on DO is pretty acceptable, but it doesn't let me sleep at night very easy.


Thanks for advance for all your input!
 

ad5ou

Active Member
Since you are a paid member, I would suggest the Advanced training class videos as your next step. That class is specifically for all of the questions you mentioned.

As for my setup, We use BDR and syncthing to keep multiple servers synced in a cluster. Then primarily rely on smart DNS for fail over scenarios. For the clients that use phones with dual registration support, they have it enabled for quicker recovery on a failure event. For other endpoints they rely on shorter registration times to find the next server if one goes down.

Luckily, so far we have only had one real world fail over event so far due to network maintenance at a data center. Everything worked as planned.
 
Since you are a paid member, I would suggest the Advanced training class videos as your next step. That class is specifically for all of the questions you mentioned.

As for my setup, We use BDR and syncthing to keep multiple servers synced in a cluster. Then primarily rely on smart DNS for fail over scenarios. For the clients that use phones with dual registration support, they have it enabled for quicker recovery on a failure event. For other endpoints they rely on shorter registration times to find the next server if one goes down.

Luckily, so far we have only had one real world fail over event so far due to network maintenance at a data center. Everything worked as planned.
Thanks for the response! I've been working my way through the online training now, there's just so many hours of video to get through, i wasn't sure if all these options would be covered, and they were sitting at the front of my mind. I was more so curious about peoples real world deployments too. Is there coverage in the video for billing as well? That's next on my list.