I am looking at installing FreePBX on a 4 host VMware 6.5 cluster.
I need a redundant hardware device (gateway) that allows use of a T1 PRI and four PSTN analog lines along with up to 8 Analog extensions that provides connectivity to the FreePBX VM. I eventually will switch the PRI out for SIP trunks in future using a Session Border Controller for security, once I have the security right. ( I have not heard of anyone hacking a PRI yet).
Do any of the SBCs do all of this? Or what device does?
How do you plan on “cliustering” these four PBX systems?
Of course if the hardware fails you will need two of them.
Tom,
The FreePBX virtual machine will live in the VMware cluster of 4 hosts. The cluster handles all of the high availability needed by the virtual machines. I have 20 virtual servers running on the cluster along with 155 TB of data space in VMware vSAN. The VMs are replicated to a 5th host at my disaster recovery site which can be spooled up if the cluster goes down. This is why I am looking for a redundant hardware solution going to the internet and PSTN/PRI
I think you’re missing the point of what I am asking. How will PBX A on Host 1 replicate to PBX B on Host 2, 3, 4 in real time? So if you add 4 IVRs and a couple of Ring Groups at 1:00PM and then at 1:20PM PBX A goes down for whatever reason and you move to PBX B on Host 2 will it have those 4 IVRs and new Ring Groups? Or will it not? Will uses still be able to get a voicemail that came in at 1:15PM but the couldn’t check it until 1:25PM, will that voicemail be on the other PBX(es)?
As well, once PBX A comes back up at 2PM how does it get all the information and data from PBX B that happened between 1:20 and 2PM?
In regards to the connectivity, a PRI and 4 POTS lines are going to require a gateway device which can support all those connections. Unfortunately, the page that @dicko supplied is garbage. 98% of the links are dead and the rest are to junk. That being said, most of those gateways requires SIP accounts that requires registration. In some cases they can peer but that will depend on the box you get. So now how does PBX A/B/C/etc, etc connect to this gateway when needed? Are the POTS there as backup for when the PRI goes down?
The next big one is, what happens if the gateway dies? Are you going to have a spare gateway on hand just in case that happens? Because that’s a serious point of failure in this scheme so far. Are you going to have a second PRI? Or just a second gateway that the POTS plug into?
Only going to have 1 instance of FreePBX supporting 120 users and 175 devices with most calls being internal. The VM can be on any one of the four hosts. If a host goes down all of the VMs on that host are quickly spooled up on another host by the High Availability option running on VMware. No data is lost. Have to test if the calls will get dropped or interrupted. The system can tolerate 1 host failure out of 4. All of this runs on a redundant UPS backed by a standby generator. The network switches in the campus all have UPS to keep the extensions alive for at least 2 hours.
As for the gateway device I plan on using one that has redundant power supplies at least. One device for the existing PRI and another for the analog POTs lines and analog extensions. The POTs lines will be the backup for the PRI and handle a couple of fire alarms.
I have several physical servers available (since I virtualized them) to use if I decide to buy the boards and roll my own. These servers are already have raid drives, redundant power supplies and network cards.
Yes I will purchase spare boards and have a cold standby server at least, HA if possible.
When I switch from the PRI to SIP trunks, running SBC software on this gateway server is an option if I do not purchase an appliance.
A pair of appliances in HA configuration that can do all of this will be considered too - which is the original question - does it exist?
To have the gateways highly available legally, you will need at least two, from that link I posted
https://www.gl.com/t1e1-multi-port-repeaters.html
For the T1 or another T1 from the provider with them doing the failover
The FXO’s can be simply wired in parallel, ,
Go to Viking for FXS multoplexors.
Of course none will work for the 5th. Host if not colocated.
Of course , most of this is readily available from your telco using call forwarding to a sip provider.
Yes, all active calls will be dropped and lost if the PBX they are on goes down. Those calls would be active in the memory/instance of Asterisk running on the current PBX. There is no way to share that or the channels across Asterisk systems.
So however long it takes for the backup host to spin up these VMs, everything to load and run is how long you will be down and unable to handle calls.
Sangoma offers a SBC with HA capabilities if you need a SBC. Also our Vega 400 4 port PRI can run in HA where you stack two of them and if one does the other takes right over.
I’m currently contemplating a high availability FreePBX setup similar to the one you’re describing (mostly because of this issue) only using KVM instead of VMware. Could you share a bit more about how you implemented (or plan to implement) your cluster?
For example, how do you detect when a host goes down and needs to be migrated? As I have seen, the PBX can be totally down while Asterisk and other process are still running. Does VMware handle floating IPs to make sure traffic gets routed to the new host after migration, or do you have a separate solution for that?
Any pointers or links you can give would be much appreciated.