CVP utilisation in Webview Trunk Usage report

Unanswered Question
Mar 14th, 2010


One of my customer recently opted for a ICM 7.2 + CVP 7(Comprehensive, SIP no Proxy) model over their existing ICM6 + IVR deployment. Since they were used to monitor IVR usage through Network Trunk Group Usage report from Webview, they are expeciting the same from new deployment as well.

Understanding the fact that CVP wudn;t directly reflect in NTG reports, I was wondering if we can do some sort of workaround and CVP ports utilisation can be fetched using NTG report itself.

Thanks in advance.


Soman Nair.

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)

The CVP Call Server licence usually activates a huge number of ports. Typically, Cisco enable 1000 Call Server ports, although this is well above the number the box can allow (maybe 750 with combined CVP and CVP VXML under SIP). So you simply don't need to monitor this in the same way that was important with the IP IVR.

Count the number of channels on your incoming T1s, E1s. That defines the maximum number of possible Call Server ports being used. If this is under the number of Call Server ports across the number of Call Servers being used (and it surely will be) then there are no worries. How do you gateways distribute the SIP INVITES across your Call Servers?

At one stage it was thought that SNMP would provide the number of Call Server ports being used, but it may not work yet. Check the MIB to see if this is part of the spec and maybe try an SNMP GET to see if you can fetch it.

You can run the diagnostic applet :7000/CVP/diag">http://:7000/CVP/diag and look at the statistics. Maybe you will see the info there.

As I said, Call Server port usage is not a normal concern unless you have a huge influx of calls - and I don't think you do.

More important is the number of CVP VXML ports - assuming you are running some Audium applications. Disregard this if you are only using microapps.

How many do you have? How are the gateways distributing the load to the CVP VXML servers? I assume you don't have a CSS, since you don't even have a SIP Proxy server pair. Must be a smallish deployment to run like that.

You can look at the call log on the CVP VXML servers - CVP\VXMLServer\logs\Call (something like that .... can't recall exactly). Whenever a new Audium application runs the number of current sessions in use is written in the logs. Take a look over a week or so and pull out this number (drop the CSV log into Excel and find the right column). How does this compare to the number of licenced VXML ports?



Chad Stachowicz Mon, 03/15/2010 - 05:42


    This strikes on a great point I have been wondering, and what is Cisco Stance on purchasing VXML Ports for Micro App Only deployments?  I have a feeling it was not Cisco's intention to not have VXML License used for Microapp's.  Has this topic ever come up between you and Cisco in a sales cycle.



You have to buy the CVP VXML ports as if you WERE using Audium apps - even if you are running all microapps.

You need to make the estimate of calls getting IVR treatment PLUS calls in the queue and buy the appropriate number. That's the A2Q requirement. Cisco don't care if you decide not to build any Audium apps.



Soman Nair Sat, 03/20/2010 - 03:39

Hi Geoff,

Sorry for this long delay, but unfortunately I couldn't get into support forum past week.

Initially there are 10 E1's (300 channels) allocated for this project and would be expanded later on. Since we are not using CSS or CUPS(Proxy), calls are individually sent to either of the CVP Call Servers.

At this moment, my goal is to provide customer with 'Inservice Time' for these trunks/CVP ports. Going for a OAMP solution doesn't sounds feasible with this customer.

I can generate a custom report, if I gt to know how much time, these ports were used for incoming calls before or after connecting to agents, i.e - even for the time my VXML Gw was handling call...

With my previous IVR setup, I was using below mentioned fields of Trunk_Group_Half_Hour table from AWDB. This is also reflecting in IVR Trunk Utilisation reports of Webview. Do/Can we have any similar data in AWDB/HDS db with CVP scenario.




Perhaps you could describe more carefully your gateway and carrier trunking arrangement.

Let us assume you have two combined (ingress and VXML) gateways with 5 x E1s in each. When the call arrives at the gateway, a dial peers sends it directly to a Call Server. I guess the dial peer on gateway A sends it to Call Server A, with a second preference dial peer that sends to Call Server B should the SIP INVITE to Call Server A fail; the dial peer on gateway B sends it to Call Server B, with a second preference dial peer for Call Server A.

Since the gateways are combined ingress and VXML, you use send to originator to run the VRU leg on the originating gateway.

Let us further assume that the carrier has these 10 E1s in two trunk groups and is alternating imcoming calls between gateway A and gateway B. This is obviously a better arrangement than having the second trunk group only used when the first is full. In this case, the absence of a SIP Proxy server to round-robin your calls between Call Server A and Call Server B makes for a very asymmetric system under varying loads.

Maybe you try for symmetry in the absence of a SIP Proxy by using two equal preference dial peers, DNS lookups, or local SRV config on the gateways.

Please correct my assumptions.

Remember that CVP is not like the IP IVR at all - the treatment and queuing is done on the gateway, and a CVP port is used for the LIFETIME of the call, even when the call is extended to the agent (assuming VoIP transfer). So when you say:

"I can generate a custom report, if I gt to know how much time, these ports were used for incoming calls before or after connecting to agents, i.e - even for the time my VXML Gw was handling call..."

I think this shows a slight misunderstanding of the mechanics of CVP.

At any point in time, the number of Call Server ports in use is the sum of calls receiving IVR treatment, calls being queued, AND calls with agents under CVP control. With the large number of CVP ports across your two Call Servers (1500) and 300 inbound channels, you are never in jeopardy. You have oodles of capacity.

The two things that really matter are:

1. The capacity of the gateway for VXML sessions and the licencing of the VXML sessions on the gateways.

2. The number of CVP VXML ports (Audium licenced ports) across your VXML server and how these are load balanced (without a CSS your only way of balancing is through the ip host table on the gateways perhaps using the routing client method described on page 12-2 of the CVP7 SRND)

What sort of gateways do you have?  Assuming you have 3845s, these support 230 (VoiceXML and DTMF) sessions - less if using ASR. Across two gateways you can handle 460 sessions - well above your 300 channels, so no worries on that score. As far as licencing goes, you need to use the IVR calculator (knowing the BHCA, the treatment time, number of agents etc) to figure out how many licences you should have bought. These are currently not checked in and out on the fly, so as long as you have provided the right info during the A2Q and purchased correctly, Cisco will OK it - bursts that exceed your licence will not be penalised.

As I said originally, 2 is really the only parameter that you would want to monitor. If you have (say) a menu system built in CVP VXML and you have used up all the licences, the next caller will be automatically queued until a licenced port becomes available. Not desirable, but not a show stopper. Monitoring the usage of these ports in real time is the only thing I would see a need for.

Perhaps you can provide the true parameters of your system and we can discuss this further.




This Discussion