cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
935
Views
4
Helpful
14
Replies

MGCP vs H.323 with CallManager

patwill66_2
Level 1
Level 1

Does anyone have a good explanation of the differences between MGCP and H.323? I have been searching online but I haven't found any real good information.

The reason why I ask is because I currently have MGCP configured on my voice router with CallManager talking to it. I would like to monitor the active voice ports on the router but I guess (per TAC) when using MGCP, the router is like a dumb terminal and doesn't do any of the call processing so it doesn't track the number of active calls. These stats are obtainable from CallManager using the Real Time Monitoring Tool but I would like to be able to poll a specific OID using SNMP and be able to graph the results historically.

TAC mentioned that I could change the protocol from MGCP to H.323 and I would have some different options as far as OID's to poll. I guess with H.323, the router actually does the call setup and teardown and can track the stats of active calls, which would give me what I am looking for. I just don't know how smart it is to change my configuration and what kind of risk I am going to incur. I read that H.323 is more difficult to configure and requires more administrative overhead, where MGCP is very straightforward. Does anyone have any insight to this?

I just want to monitor active calls using SNMP, it cant be that hard…

14 Replies 14

Avner Izhar
Level 3
Level 3

If it is a PRI interface that you are trying to monitor, I think there is an OID which can show the amount of used b-channels. Wouldn't that be a good information? most calls are probably pots to voip, right?

Anyways, mgcp vs h.323 is like you said. An h.323 gateway is an independent device, it has it's own dial-plan and require a separate configuration. On the other hand, if you already have srst on your mgcp router, then you are already configured for h.323.

HTH, Avner.

I found the MIB for ISDN channel status (CISCO-POP-MGMT-MIB) but this didnt return any good results. Almost everything had a value of 0 when I walked the table. The calls are POTS to VoIP, yes, but for whatever reason this did not work. It is looking like switching over to H.323 might be the only way to go.

Will test it today and would let you know if I can find another OID, should be one that you can use.

Well, the best thing I found was this mib:

Object: callActiveCallOrigin

OID: 1.3.6.1.2.1.10.21.1.3.1.1.10

Type: INTEGER

Values 1: originate

2: answer

3: callback

This one will generate an entry per call leg when you walk it, so divide the results by two and you have the number of active calls.

HTH, Avner.

This may work for now. I am still going to have to work on the H.323 option if I want a truely supported way of polling a MIB for active call information. Thanks for the help.

paolo bevilacqua
Hall of Fame
Hall of Fame

And don't forget SIP, that is almost the same to configure as H.323.

There is really no risk using either, unless you want unsupported things like caller-id with MGCP on T1 CAS, etc.

Is it the case that early versions of CCM only supported a certain number of MGCP gateways, whereas this limitation did not apply to H323 gateways?

In terms of making a decision between the two - this could be quite important factor to consider.

Ajaz

No, that is not the case. I only have three gateways configured in CallManager. This would not be a factor for me. The only factor I have to consider is the amount of work it will be to reconfigure the router for H.323. My company runs 24/7 so I can only have a couple of minutes for downtime, and since I have never converted a config from MGCP to H.323, this is the part that worries me.

Don't forget that MGCP will support caller-id on a PRI; whereas, an H323 gateway will not. An H323 gateway will support caller-id on analog lines (FBs); whereas, an MGCP gateway will not.

It is easier to configure toll-bypass when using MGCP because the call is controlled by CCM. CCM can also pull a call back from an MGCP gateway if resources aren't available. This can be done with H323, but it takes a little bit more configuration work on the gateway.

When you say that caller ID is not supported, what exactly do you mean? Inbound calls to my office we only get the digits of the person calling in. Howver, if calling outbound to a location that has call ID, that location will get our number and our name. So is it just the name that wont show up or the ANI as well?

In fact, H.323 is able to deliver both calling number and name.

Then, if your telco delivers a calling name, that is a different issue, and you have to verify that with telco,

Traditionally H.323 didn't support calling name, but I hear this was fixed in a recent IOS update. Does anyone have the specifics? I know it didn't work in 12.4(11)T3

Just thought I'd throw in our experience with MGCP vs. H.323. :)

We use MGCP for the location of the Call Manager and H.323 for our remote offices that are on our Call Manager. I can't remember if I read it or TAC told me, but this is considered best practice because of SRST. MGCP is much easier to configure since the Call Manager is in control for the most part. I highly advise getting TAC involved when configuring H.323 because there is always something you'll overlook and we've encountered some bug everytime.

Be aware that each PRI connecting to MGCP needs it's own d-channel.

One of the reasons I've not been a big fan of MGCP is I once saw a really nasty outage that wouldn't have happened with H.323. A phone technician was working in the wiring closet, and his tool belt got suck on some stray wires for one of the PRI lines. The voice gateway correctly saw the D-Channel drop and the line was marked bad, but nobody noticed since calls simply failed over to the other ones. A few weeks later, I reset the MGCP connection and it never came back. After several hours of troubleshooting with TAC and the Telco, I physically unplugged the damaged PRI and everything popped back up.

You can certainly blame poor cabling for that, but still I found it really scary that one bad PRI hosed the entire MGCP backhaul.

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: