We have a cisco Voice implementation that I am trying to understand the components.
We have two IPCC servers, one that handles calls from outside and these calls are directed to the internal users in our company.
The other IPCC server handles calls from outside callers to our Voice IVR system.
I have some general questions about IPCC:
What exactly does IPCC do?
Does Call manager just handle call routing on a very basic level?
I know there are scripts created and the IPCC server houses these scripts.
How does Call Manager fit into the inbound call routing from outside callers?
Does Call Manager direct the calls to the different IPCC servers depending on what the user is looking to do?
IPCC is a contact center solution with IVR capabilities. As you said CallManager is reposibile only for redirecting calls to devices, plus typical PBX functionallity. In order to be able to perform any advanced routing, digit collection, playing prompts, routing based on area code, etc you need applications such as IPCC.
CallManager directs calls to IPCC system based on called number, so for example if someone calls a paricular DID or a toll-free number the call can be redirected to IPCC for advanced treatment. without seeming your IPCC scripts I cannot speculate what exactly your system is doing, and what is the reasoning behind 2 seperate IPCCs.
How does CallManager fit into the inbound call routing?
Call from the PSTN goes into the Voice Gateway. Voice gateway sends the call to CallManager. I'll assume you have an ISDN circuit. CallManager analyzes the callED number (also called DNIS).
It will match a CTI Route Point in CallManager.
The CTI RP is associated with a CTI-enabled JTAPI user in Global Directory in CallManager. That JTAPI user account belongs to one of your IPCC servers and sends the call over JTAPI to the appropriate IPCC server.
Now we are in IPCC. We still know the callED (DNIS) number and it matches a JTAPI Trigger. The trigger is associated to an IPCC application. The application calls a script and the script runs.
The rest depends on what the script is doing.
So CallManager will direct the call to the IPCC server A or server B by having two JTAPI/CTI users in CallManager's global directory. One user belongs to IPCC server A. The other belongs to IPCC server B.
The first JTAPI user associates (owns) certain CTI Route points and CTI ports that "belong" to one of the IPCC servers.
Same for the other.
Thank you for the excellent replys from both of you.
This brings up more questions:
"That JTAPI user account belongs to one of your IPCC servers and sends the call over JTAPI to the appropriate IPCC server."
1. Can you explain "sending the call over JTAPI"
I understand the JTAPI trigger for us is a four digit number used to route the calls.
2. Refering to the attached Voice GW config, the blue shows the dial peers configured which have the "session targets" configured which are our Call Manager servers. So my understanding is that the call gets relayed to the GW and the destination pattern is read and that matching call pattern relays the call to the "session target" or Call Manager, correct?
3. The dial peer in red is from the Nortel system, which is taking inbound calls from our outside callers and sending them to the Voice Gateway, the nortel is giving them the DNIS of 1000, which then sends them to our Call Managaer, sound correct?
4. I do not know what the purple is doing, unless the nortel is also matching a destination pattern, giving those calls a DNIS of 6520 and sending to CM.
5. One of the IPCC servers is an IVR with TTS, the two IPCC servers are configured as an HA cluster, with a cold standby in our DR site. My understanding is that the Call Manager servers have LDAP configured that the IPCC servers are pointing to. The idea is that if the two HA servers were to fail, the cold standby could be brought online and it will check the LDAP database for any changes and everything will be fine.
Does this sound feasable?
1. JTAPI is a programming "conduit" that CallManager and IPCC know about. CCM and IPCC can pass information between one another through this JTAPI link. JTAPI is also referred to as an Application Programming Interface (API). APIs help programmers of software more easily write programs and can provide commonality, portability to other operating systems, etc.
2. Yes. Your subscriber as the first choice (preference 1) and to the publisher if the subscriber was down (preference 2)
3. Inbound calls go to the gateway. Hit dial-peer 160 because it has the "incoming called-number" that matches the 1000 that Nortel sent. Then matches dial-peer 300 as the outbound dial peer and sends off to your CallManager subscriber. Note, if your subscriber were to go down, 1000 would get a fast busy. You need your backup dial-peer.
4. The purple is another outbound dial-peer going to CallManager. The incoming dial-peer is again the dial-peer 160. That is your only incoming dial-peer. Can't tell if the call comes from AT&T or Nortel. You would have to do some debugs.
5. So you have three IPCC servers? Is this IPCC Enterprise or IPCC express? And what version? I can't tell if it sounds feasible or not at this point until we get more information. Once the Enterprise vs Express and version is known, we will need to know some other things.
One is how you are doing your backups and what kind (BARS, drive swap, 3rd part full backup, etc.)
Wow, thanks for the great replys.
Can you tell me what you see that ties the "incoming called number" to 1000 and 300.
I am missing that.
I believe that the IPCC is express, where do I fing the versions?
That's the problem. I can't.
You have one dial-peer that is set as an incoming dial peer. That is dial-peer 160 pots. That incoming dial-peer allows any match (the dot-T). Since there are two trunkgroups assigned (the ATT and the Nortel), the 1000 and 300 can come from either.
Remember, every call on an h.323 router has an incoming dial peer and an outgoing dial peer.
To determine what you have, (this will be for Express.) Go into Appadmin and it should show you on the main screen.
What components of the config point out the two differnet trunk groups?
What is h323?
We are on the verge of implementing a new design from Verizon.
This design will be using new 3845 routers as our voice gateways.
The calls will be hitting the 3845s as g729 and it will be converted to g711 on our inside network.
My understanding is that it will be cheaper to use 729 as we get the calls from Verizon.
The 3845 will house DSP cards to convert the calls from 729 to 711.
Why is it cheaper for Verizon to send the calls as g729?
If it is cheaper, why not just leave it as 727 on the internal network?
What is a DSP?
What is the type of service you are getting from Verizon?
Is it going to be SIP trunk?
You are talking about transcoding from one codec to another,so I am assuming the 3845 router will be an IPIP GW as oppose to PSTN gateway, am I correct?
The only benefit of G729 over G711 is that it uses a lot less bandwidth (30kbps vs 87kbps over ethernet). However G729 requires more horsepower to compress, thus more DSPs on the Gateway. Quality wise G711 is better, but G729 is good for speech (not so much for Music).
DSP is a digital signal processor, basically a chip in the router (as well as phone) that allows for digitalzing voice.
H.323 is a voice over IP protocol which allow for controlling a voice GW, there are variations of VoIP protocols: H.323, MGCP, SCCP, SIP, etc, each with it's own pros and cons. In order for a CallManager to talk to a voice gateway one of the voip protocols needs to be used.
Your dial-peer 160:
dial-peer voice 160 pots
description Inbound calls from ATT and Nortel
incoming called-number .T
the two trunkgroup commands. One points to ATT. The other to NORTEL.
Now we look elsewhere in the config for the definitions. Towards the top, we have:
trunk group NORTEL
trunk group ATT
There are no commands under there so they take defaults (hunt order, etc.)
We have to assign voice ports to a trunkgroup. For analog (you not assigning any), you would find a trunkgroup command under voice-port x/y/z. For T1's, we put them under the controller interfaces. So.
controller t1 1/0
controller t1 1/1
controller t1 3/0
Go to your PBX. And you will probably find three cables leading from the voice gateway to the PBX.
controller t1 3/1 goes to ATT. And you should find a cable going from the router to an ATT circuit.
Cderen did a nice job of answering your other questions.
You guys are a great source od information.
If you want me to open another post I will, but have this question:
I have a failover implementation that will be used for our DR site. There will be two 3845 routers, one to bring in a new DS3 for voice, another to use IP to IP gateway.
The DR site has 8 DIDs on three PRIs to use during our DR tests and in case of a real disaster.
When we implement the DS3, the PRIs will go away and I will need to port the DIDs for use.
I was thinking of using the 4 slots on the 3845 and installing 4 2port FXS WIC cards.
Will this work?
I don't know exactly what you need, but yes, it would work. So won't 2x VIC-4FXS/DID cards.
Problem is, any one of the DID's will only be able to take ONE call, since an analog connection can only take one call. Is this what you want?
Why now just keep a PRI?
Thanks for the reply.
Yes this is what we want.
The three PRIs are going away and the DS3 will take it's place.
We will port over the DID numbers from the PRI to the DS3.
We only need the DID numbers in the event that there is a total disaster and outside vendors need to use a modem to get to our equipment.
We also have modems connected to a couple of workstations for dial out.
I found out the FXS ports are not needed.
But to answer about the DS3, Verizon is going to terminate the voice DS3 to a 3845 router to a NM-1T/E3 module, nothing really unusual as far as I know.
Verizon ordered the original 3845 routers for the HQ site and they are up and working, we have not cut our circuits over yet though.
I'm pretty sure that module doesn't do voice. I don't think that module can even section off a number of channels to do voice.
So the carrier would have to terminate the T3 onto their equipment and section off X number of channels for voice. They will provide you an RJ45 for an ISDN PRI and perhaps a coax for the remaining T3.
You will have to put the PRI in the appropriate VWIC module (say VWIC2-1MFT-T1/E1) with appropriate DSPs and IOS feature set.
No data over the DS3 only Voice nad I believe they are using SIP.
I don't understand it but,
I believe I heard mention of SIP.
There are two routers, the DS3 terminates to one and the other will have IP to IP gateway installed.
I just checked the DS3 router and the cial peers are showing "session protocol sipv2".
How does the transition from the DS3 to Call Manager take place?
Does the DS3 router send the calls to the IP to IP gateway router in SIP and the IP to IP gateway translate to something the Call Manager can understand?
Also if you see this, check the "Voice Gateway, dial peer and Call Manager" post
Post both configs.
Hide passwords and any public IP addresses (hopefully you shouldn't have any.) Private IP addresses are ok to post if you are comfortable. You can always so a search and replace all to protect the innocent.
But I need to see the IP addresses to know how the call will get from one router to the other and to CallManager.
Will need your CallManager IP's also to determine how the call gets there.
Well, they have not been configured just yet.
Originally this was configured differnetly and we are going through a redesign, so 90% of this has not been done yet.
I was just curoius how it is usually done. The way it is being proposed, it looke like the IP2IP gateway router will be all IP in the ethernet LAN, but not sure.
IF the DS3 termination is done to the LAN, then I am thinking the IP2IP gateway would be a single Gigabit ethernet link to the etherent of the DS3 gateway.
Origianlly the config (the way it is now) is the DS3 router is receiving from Verizon and has 6 T1 cards. They are going to a CMM blade in a 6509 switch. Call Manager is also on this switch.
You should probably have these discussions at this point with your design team. There could be 100 ways why they are doing this and that.
They may have used the IP-IP GW on another router (it's called Cisco Unified Border Element now --CUBE) because the DS3 router will be busy handling that DS3.
I'm confused on your statement with the DS3 router and the CMM module. The PRIs should go directly into the CMM module.
Feel free to keep asking questions, I'm just not sure, because I can't see all the detail, that I would give you a 100% correct answer.
Here is some new information on this and see if I am making any sense (keep in mind I know nothing of voice):
Inbound calls will traverse the Verizon SIP trunk on a DS3 and be terminated on our 3845 edge router.
There will be three virtual trunks created by verizon:
Normal non customer calls (local and Long Distance)
Customer Calls Local
Customer Calls Long Distance
All calls will be compressed and getting to us as g711 calls.
We will be sending all of this traffic to our IP2IP Gateway router (second 3845)over Ethernet using h.323. The IP2IP Gateway will have DSP resources to transcode only the Customer calls to g729, which I think will then go to Call Manager ( again h323).
I believe we will be using the CMM module in the switch to transcode the rest of the calls (not sure about this part, but I think I heard someone mention this), then send them to Call Manager vis h323.
I am sure about having to use the CCM blade, but not sure why. No PRIs will go to the CMM blade. Using IP2IP gateway was suggested by Verizon to allow much greater throuput rather than having PRIs go from the 3845 to the CMM blade
Why not just have the CMM blade (if it is being used to Transcode) transcode all the calls, or have the router transcode all the calls?
Why split it up this way?
I am sure that the Customer calls will all be using the IP2IP Gateway router to transcode those to g729 with the router's onboard DSPs, these calls are also utilizing Text to Speech, would this be a reason?