I have two Asterisk servers, one with 4x E1 (server A) and another with 16x E1 (server B), both running Asterisk 10.12.2, DAHDI 2.6.2 and libss7-1.0.2. I have recently set up TDMoE between the two servers and DACS to map four of the physical E1's on server B to virtual E1's on server A. Two of the four physical E1's on server A are running signalling links and the remaining two physical E1's and four virtual E1's are running only voice circuits. This appears to be working 100%. I have not yet tried running a SS7 signalling channel over the TDMoE+DACS E1's... In theory it should work (the voice circuits work just fine). Has anyone done anything similar, perhaps on a larger scale? My thinking is to use this to try and 'cluster' libss7. I realise that it isn't a true clustering solution (if the master server fails, the entire solution fails), however, my hope is to reliably run 16x physical E1's on four different slave servers and map these (via TDMoE and DACS) to 64 virtual E1's on a master server that then does signalling and should, hopefully, be able to handle the lower interrupt demands of the TDMoE spans. The one issue I am encountering so far, however, is that the hardware echo cancellation isn't being used on the physical E1's with DACS mapping to TDMoE E1's. I don't really want to be doing software echo cancellation for 64x E1's... In the test setup I have not installed hardware transcoding boards, however, this would probably be necessary at that scale. At least that is possible. Hardware echo cancellation doesn't seem to be possible for a channel received on TDMoE.
|