All of my CVP installs have been either H.323 or SIP with without CUPS. So, I have a question.
On the initial setup, when configuring CUPS with the static routes, I am confused on what I do for the static route for the following components:
- Labels - If my cvp label is 1111111111, I'll create a static route of 111* but where do I point that to? It says to point it to the VXML Gateway, but in a large branch environment I may have 100 voice gateways. Do I pick a few gateways to use for IP originated calls and alow SIP to always use the originating gateway?
- Same thing goes for the error and ringback. lets say they are the default of 91919191 and 92929292. Where do I point these to as well in a large branch deployment?
Solved! Go to Solution.
As I understand it, you still need to use the "sendtooriginator" settings in the CVP SIP service to ensure calls are sent back to branch gateways.
Corey is correct.
Configure your NVRU label in the "Send To Originator" section of the SIP config on the Call Servers and save and deploy. Likewise, configure the error label and the ringtone label in the "Send To Originator" area.
This will ensure correct functionality in a branch office deployment where the ingress gateway and VXML gateway are one and the same. The VRU leg will be executed on the same gateway that starts the switch leg.
Calls that originate from inside the network (from IP phones) are much harder.
With H.323 there is a nice (but somewhat complicated) way to do this, so that the call is forced back to a gateway local to the IP phone (at the same branch).
In SIP, the picture is less clear. One can create multiple customers and arrange NVRU labels appropriateley with the SIP Proxy sending those labels to branch gateways, but this leads to a more complex configuration.
I'd love to know a better way.
Here is another twist to this. If you have seperate VXML and Ingress gateways, the "send to originator" does not work.
In this case, do I just create static routes of 92929292 and 91919191 in CUPS for example and just use the same priorities? This way at least it will load balance across all three. I plan to do this for all the Dialed Numbers as well.
I have three AS54xx's that are dedicated VXML gateways so for all Dialed Numbers, I'll just have three static routes in CUPS with the same priority.
I do just that, my ringer and error.wav load balance over 2 gateways.
pri = 1
weight = 50 on both static routes for ringer and error.
I am doing all my VXML Gateway / Ingress gateway preference via sip.sigdigits and priritized SIP Static Routes in CUPS. I just use the SendtoVRU Label. I guess the Send to originator only needs to be used because its less management then Sig digits? I guess as andy puts in his post about having seperate VXML and Ingress gateways, the sig digits also allows you to have split VXML / Ingess Gateways and still have perfect prefernced WAN routing, we have this exact setup of one ingress and one VMXL Gateway at each location. The downfall is that SIP Proxy redundancy is done through DNS SRV and the best you can do is load balance into the proxies. Ick.. I would imagine using 3 sig digits you could accomodate about 900 remote sites using the SendtoVRU node. Maybe send to originator is easier though. I also took all inbound calls into my CCM's through the CUPS Server so I needed Sig Digits to optimize normal call routing aswell for 4 sites. Outbound I did all calls in a central SIP Dial plan. I utilized a site-code type feel in the SIP Proxy. This also gave me the ability to let them easily expand their TEHO to 100 sites accross the USA using only 1 'sitecode for LD' and making all sites route pattern go to this same site code. TEHO is gorgeous :)
Here I am attaching some configs of a VXML Gateway and Ingress Gateway.
so All dialed number's in ICM are 6 digits starting with 70xxxx
so calls come into my ingress as 4 digits, my translations patterns matchs the dnis, expand it to 7 digis (1st digit is a sig digits used for wan preferenced routing) and send it to the sip proxy. Sip proxy looks at the sig digit +70 and determines which call server to send it to. The call server has sip.SigDigits = 1 set in the sip.properties file on the call server, this field cannot be set through OAMP. coming in 7.1. rom here it goes into ICM. Now under your Network CRU put the SAME label for each CVP Routing Client. When it goes back out the call server will prepend whatever sig digit it stripped. it then send sigdigit + 6 digit VRU (my config 100050) + orrelation ID to sip proxy. From here just use the sig digit to route it in the sip proxy to the correct VXML gateway First.
Let me know if you haev any other questions.
Interesting idear' chad!! Way to think outside the box. Most of my installs, Send to Originator would be just fine but I can already think of a few intances where this would have helped.
No worries this design was about a week of battle between our Solutions Architect, Myself and one of Our IOS/CCM Gurus. Brains combined we came up with this and it works great!