[cisco-voip] DTMF interworking on CUBE - asymmetric payloads

Tue Jul 19 07:51:07 EDT 2016

Thanks Justin. My concern is with nte's coming in a mix of different
payload types depending on the end point. Since it seems I can only specify
one payload type for them, when I get an offer using a different value,
CUBE doesn't include it in the answer. And then no interworking happens
since we didnt' successfully negotiate a DTMF relay method at all. I'm
hoping with the asymmteric payload feature that cube will negotiate and
pass through the dynamic payload type and also actually look at the packets
and interwork them to kpml. The more I think about it the less likely it
sounds, but I may as well try it.

I think what I really need for it to do is just negotiate EITHER 101 or 98
for dtmf relay out to the 3rd party, but i'm stumped on if/how to make that
work.

On Mon, Jul 18, 2016 at 9:21 PM, Justin Steinberg <jsteinberg at gmail.com>
wrote:

> yes, CUBE can do RFC2833/NTP to a Telco and SIP-KPML to CUCM.   I do this
> for calls that terminate on CCX IVR since CCX does not support RFC2833.
> With only rtp-nte on the dialpeer from CUBE to CUCM, CUCM will invoke a
> MTP.   Adding sip-kpml to the dial-peer will allow RTP directly from CUBE
> to CCX without any MTP in the middle.
>
> On Mon, Jul 18, 2016 at 5:08 PM, Ed Leatherman <ealeatherman at gmail.com>
> wrote:
>
>> Thanks Daniel, that helps a lot in understanding the feature. I'm curious
>> if CUBE will also translate digits to KPML in this case if the leg to CUCM
>> has that negotiated. Wish I had a lab built out for this :)
>>
>>
>>
>> On Mon, Jul 18, 2016 at 4:22 PM, Daniel Pagan <dpagan at fidelus.com> wrote:
>>
>>> Ed:
>>>
>>>
>>>
>>> I specifically worked with the dynamic payload option for a few cases
>>> that came my way. Based on my findings, when a dynamic payload type (such
>>> as 100/101/etc.) is received by CUBE, it will check if the next-hop
>>> dial-peer has the asymmetric payload feature enabled and, if it is, will
>>> pass the received payload type through to the next call-leg. Take a look at
>>> my screen shot below. This was taken from some old notes where AT&T was the
>>> customer’s carrier.
>>>
>>>
>>>
>>>
>>>
>>> The call flow above shows two call-legs, and *the arrows represent an
>>> offer/answer exchange*.
>>>
>>>
>>>
>>> With asymmetric payload enabled on both call legs, the 100 offer from
>>> ATT is passed to CUCM despite 101 being the default PT for NTE. In the SDP
>>> answer from CUCM, we’re getting PT 101 -- since asymmetry is enabled on the
>>> DP to ATT in this call flow, we pass the 101 through to ATT despite having
>>> received PT 100.
>>>
>>>
>>>
>>> This results in asymmetry on our negotiated PT for each call-leg.
>>>
>>>
>>>
>>> *Let’s change it up a bit… A second example.*
>>>
>>> If asymmetry was disabled on the dial-peer to CUCM but enabled to ATT,
>>> we would receive 100 PT from ATT, send 101 to CUCM, receive 101 from CUCM,
>>> and send 101 to ATT. The resulting PTs would be symmetrical between CUBE
>>> and CUCM, but asymmetrical between CUBE and ATT.
>>>
>>>
>>>
>>> See screenshot below for a third example:
>>>
>>>
>>>
>>>
>>>
>>> This example shows asymmetric payload disabled on both call-legs using
>>> the same call flow. CUBE receives PT of 100 from ATT -- the outbound
>>> dialpeer has asymmetry disabled, so it transmits the PT specified for that
>>> dial-peer (default 101 or any hardcoded dynamic PT) to CUCM. We then
>>> receive 101 from CUCM and, since our inbound dial-peer has asymmetry
>>> disabled, CUBE sends 100 to match the original PT it received. Asymmetry is
>>> disabled so CUBE is not passing the received dynamic PT through to the
>>> next-hop dial-peer - we have symmetry on both call legs for our NTE PT.
>>>
>>>
>>>
>>> Note that CUBE has no issues receiving one dynamic PT for NTE and
>>> sending another (ex: receiving PT 100 and transmitting 101 for RTP-NTE) on
>>> the same call leg.
>>>
>>>
>>>
>>> Hope this helps
>>>
>>>
>>>
>>> - Dan
>>>
>>> --------end attach---------
>>>
>>>
>>>
>>> *From:* cisco-voip [mailto:cisco-voip-bounces at puck.nether.net] *On
>>> Behalf Of *Ed Leatherman
>>> *Sent:* Monday, July 18, 2016 3:10 PM
>>> *To:* Cisco VOIP <cisco-voip at puck.nether.net>
>>> *Subject:* [cisco-voip] DTMF interworking on CUBE - asymmetric payloads
>>>
>>>
>>>
>>> I'm trying to get my head wrapped around some DTMF interworking
>>>  features...
>>>
>>>
>>>
>>> I have this setup:
>>>
>>>
>>>
>>> UCM ------ CUBE ------- 3rd party system
>>>
>>>
>>>
>>> For both call legs through CUBE I'm advertising kpml and rtp-nte for
>>> dtmf-relay
>>>
>>>
>>>
>>> The 3rd party sometimes sends me rtp payload type 101 for nte's, and no
>>> kpml, and things work (as a bonus I observed CUBE correctly interworking
>>> the nte's from the pbx into KPML, so uccx didn't break).
>>>
>>> Sometimes they send type 98 and no kpml, and things don't work.
>>>
>>>
>>>
>>> I'm trying to understand what is happening and what feature should fix
>>> it (without breaking other things)
>>>
>>>
>>>
>>> Assumption:
>>>
>>> "dtmf-relay rtp-nte kpml" is telling CUBE to offer/accept rtp type 101
>>> only for nte. I observe that CUBE negotiates KPML only for the associated
>>> call leg back to UCM and doesn't bother with rtp-nte, so its just like any
>>> other codec that CUBE doesn't care about.
>>>
>>>
>>>
>>> So.. if third party system ONLY sent me dtmf-relay with payload type 98,
>>> could I just set the rtp payload type for this to 98 on the inbound dial
>>> peer? would CUBE then correctly switch these up to 101 headed back to UCM?
>>>
>>>
>>>
>>> How can I (or can I at all) make this work in my particular case were I
>>> could receive both?
>>>
>>> I see "asymmetric payload dtmf" thrown about as a possible solution, but
>>> I'm having trouble understanding what it actually does. It sounds like it
>>> passes these payload types through CUBE, so UCM could be getting rtp
>>> payload type 98 - it knows what to do with it? It seems like then CUBE
>>> wouldn't be able to translate things to KPML this way...
>>>
>>>
>>>
>>> I'm reading
>>> http://www.cisco.com/c/en/us/td/docs/ios-xml/ios/voice/cube/configuration/cube-book/voi-dymc-payld-dtmf.html
>>> but I guess I'm just not drinking enough coffee today (or too much) and I'm
>>> not getting what exactly this command does.
>>>
>>>
>>>
>>> Anyone know how that asymmeteric command works?
>>>
>>>
>>>
>>> --
>>>
>>> Ed Leatherman
>>>
>>
>>
>>
>> --
>> Ed Leatherman
>>
>> _______________________________________________
>> cisco-voip mailing list
>> cisco-voip at puck.nether.net
>> https://puck.nether.net/mailman/listinfo/cisco-voip
>>
>>
>

-- 
Ed Leatherman
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/cisco-voip/attachments/20160719/1e744ccb/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.png
Type: image/png
Size: 2755 bytes
Desc: not available
URL: <https://puck.nether.net/pipermail/cisco-voip/attachments/20160719/1e744ccb/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image003.png
Type: image/png
Size: 8218 bytes
Desc: not available
URL: <https://puck.nether.net/pipermail/cisco-voip/attachments/20160719/1e744ccb/attachment-0001.png>