Customer Virtual Server & Trunk Server

We’re moving towards hosting clients.

I’m looking at setting up a virtual server to each client, and using a physical box to terminate their land lines at.

At this moment we don’t have a T1, just POTS lines. Normally we would restrict the client to use only 1 or 2 channels of the T1, but for now with POTS, is there a way we could restrict the client to use only their dahdi channel?

That would be a no for PRI T1’s, for older 24 channel robbed bit e&m T1’s (super trunks) then you can map channels as needed (dahdi/1 to dahdi/24 in the first span) for outbound calls. But you will have a problem with inbound DID’s if you use them as they will generally come in hunting ascending, thus denying those clients using the lower channels when in use.

Re-reading your post, if you want to that from one hardware machine to several “virtual” machines, then also yes with e&m for outbound, you use TDMoE (dynamic eth) between an instance of dahdi running on the hardware box and a mapped passthrough to several similar dahdi TDMoE trunks running on the virtual machines, cross-connect the relevant channels as appropriate in the hardware box dahdi to it’s relevant dynamic TDMoE spans connected to the VM’s , you don’t need asterisk running on that hardware box, just dahdi. TDMoE is of course a layer two protocol and not tolerant of packet delay/loss , so you will need everything to be on a very well connected ethernet segment.

And of course you can also do that with ANY dahdi compliant hardware on the Host machine, including FXO’s

I’ve done this before with a sip or an iax trunk between. I just lost my notes and forget how!

Once I get it figured out again i’ll post it up. It worked slick when we did it before.

Try the TDMoE solution, it works slicker and you don’t need asterisk on your host machine.

I got it working, but yours does sound interesting.

Any suggested reading to learn about it?

Actually it’s all in the sample system.conf file that comes from Digium, you are just redirecting 1-to-1 each 64k TDM/analog channel across the layer 2 “spans” between the machines using the mac addresses of the networks chosen, on the master that is the dacs/dacsrbs stuff, on the slaves, you use fxs/fxo/e&m whatever, the passed-through signalling is transparent to the slave onto the underlying dahdi hardware on the master. The “timing” on the slaves is derived from the master, the spans/trunks between the machines can be any number of channels wide.

Thanks! i’ve got some reading to do.

My bigger problem is this.

We normally use analog lines, we setup no answer call forward and busy call forward so when the clients main number is used, it will bounce over to a voip number.

How could I do this with a DID on a T1?

I personally would do that with a true proxy running on the master (think kamailio/freeswitch for the TDM bit), but depending on scale you can have your telco hunt T1’s and add “common blocks” to the circuit(s) giving you multiple appearances of the old single channel PSTN’s, if you call forward busy you will need to add additional call paths to that forwarding.

For delivering failsafe Fax/postagemachines/tivos FXO type services to underlying machines in a virtualized environment it works very well, as it does for transitioning legacy PBX’s to voip.