[Asterisk-Users] * point to point t1 solution?

[Andrew Kohlsmith]

On Friday 27 January 2006 15:56, Damon Estep wrote:
> that was the goal - a end to end TDMoE path terminated at both ends at
> the t1 ports on the Digium card - aware of the idle bandwidth
> requirements (about 2.0mbps including packet overhead).

I don't think it's possible to get "no signaling awareness" with "idle 
bandwidth awareness" -- they don't mix.  You either deterimine idle channels 
by watching the signaling (CAS or CCS) or you don't and pass everything 
unmolested, including idle channel "data".

> I have not seen any information that says that it is possible to nail a
> t1 port to a TDMoE channel group on one end and vice versa so the
> signaling could be passed unmodified.

Admission: I have never used TDMoE, but I understand telephony and the concept 
behind TDMoE and TDMoIP very well.

TDMoE just takes the 1544000 bits per second (192 bits of 24 channel data + 1 
bit of T1 frame data, send 8000 times per second) that a T1 produces and 
encapsulates it in raw ethernet frames.  This is going from memory but I 
believe it takes the 24 timeslots (24 bytes) * 8 plus the 8 framing bits (+1 
byte) and stuffs it into an ethernet frame and sends it on its way.  That 
would certainly give you 1000pps as someone else mentioned.

Again (from memory, and from looking at ztd-eth driver it looks to be right) 
there is no interpretation or manipulation of the signaling data done at all.   
The bits are passed as-is, which means if you're using the MCDN/NAPN/SL1 
protocol between Norstar systems that it should all work.

> You have to decode the signaling, pass the media, and reproduce the
> signaling via asterisk at the remote end, correct? This means that id
> the signaling IS non-standard it is a no go.

Only if you're trying to convert the T1 channel data to SIP or otherwise 
manipulate the bits.  If you're only trying to get them to go from A to B 
then no, no interpretation is needed.

If you're looking for a straight T1-ethernet bridge, TDMoE should do the trick 
for you.  If your microwave link appears as an ethernet bridge (not an IP 
router) and it's robust enough and has enough excess (or the ability to 
prioritize ethernet frames based on MAC address) then this should work.

If your microwave link is indeed routing you should still be able to get it to 
work if you do some fancy tunneling.  Again, not impossible but you really 
need to do some testing.  

Not having done this specific implementation before I of course cannot make 
any guarantees.

> If it could be done, signaling matters none - the two connected devices
> would still understand each others foreign language (signaling) and the
> asterisk boxes would just provide a pure Nx64 data path.

Exactly correct, and as I said several times in this email I am not aware of 
TDMoE caring one iota about what the bits represent.  It merely passed them 
from A to B.

It's certainly an interesting application.

-A.