cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
678
Views
0
Helpful
6
Replies

Payload size and RTCP message problems with CM

Donald Loyer
Level 4
Level 4

We are upgrading our phone network to Cisco AVVID. We need to interate our AVVID platform during the upgrade with a Nortel BCM. Our only option will be using H323 for interoperability.

I have supplied a message from a Nortel Eng. as to why the 2 will not work. Are there any Cisco H323 users that would like to add there 2 cents. He seems to point to the Call Manager as the problem.

1) Payload size

Problem: During H.245 capabilities exchange, BCM announces that it supports G.729 with a 30 ms payload size and Cisco announces G.729 with 60 ms payload. After that, BCM opens a logical channel with a 30 ms payload size (the biggest size that fits both) but Cisco opens a logical channel with a 20 ms payload size.

Although one can argue that this does not violate the standards (since 20 is smaller than 30), it does conflict with BCM's assumptions (and the Nortel H.323 Interoperability Specification). That is why BCM treats this situation as an error.

2) RTCP Address

Problem: From what I have seen in the Sniffer traces, there seems to be a bug in the Cisco Call Manager. The standard (section 6.2.8.2 of H323v2) specifies that the RTCP address+port in OLC for one channel must be the same as the RTCP address+port in the OLC-ACK for the other channel. However, what I observed in the Sniffer traces is:

CiscoCM to BCM

<---OLC mediaControlChannel=10.100.118.5:28001---

---OLC mediaControlChannel=10.100.118.4:4001---->

---OLC-ACK mediaChannel=10.100.118.21:27856,

mediaControlChannel=10.100.118.21:27857------>

Note that the trace shows that the Cisco Call Manager sends OLC before it has talked to the IP set. That probably explains why the parameter in the OLC is incorrect.

Preferred solution: Fix Cisco Call Manager so that it sends the same RTCP addresses for both logical channels.

6 Replies 6

a-vazquez
Level 6
Level 6

This seems to be a bug in the call manager. For more details please check the following bugs:

CSCdw50469

CSCdv57052

CSCdw67665

And this bug has been fixed in the latest verison.

cstephens
Level 1
Level 1

FYI,

We opened the same case with TAC to no avail. The TAC engineer refused to escalate to the DE's. It HAS NOT been resolved with the latest CM 3.2.X release, but I will let you know 3.3X goes, I am downloading as we speak. I will update the forum with the results.

(So much for standards)... FYI, the bug's are not yet published, so I was not able to confirm...

Cory

I'm not sure how you are downloading CM3.3, because the only thing out there is Support Patch A for 3.3(2) which requires that you already have installed CCM 3.3(2) which can only be obtained via CD.

In any case, those 3 bug records don't seem to apply: 2 of them are specific to the 7935 IP phone and one is specific to the DT-24+ gateway, but none of them have to do with how CallManager performs H.245 negotiation.

Just out of curiosity, have either of you attempted to configure Fast Start on the CallManager side, assuming of course the BCM supports it? In the latest 3.2 software, there is a CallManager service parameter called "H323FastStartInbound" and like it says applies to inbound only. Also, I wonder if the problems you have reported apply to inbound calls, outbound calls, or both. Additionally, have you insured that the gateway types have been configured as H.225 and not Intercluster Trunk?

On Call Manager why do you not set the G729 payload size to 30 in the Service Parameters. I have used this often to reduce the overhead on WAN.

You will of course have to set all other H323 devices (Including IOS H323 Gateways) to use 30 byte sample size . Otherwise the packet size will not match in each direction

We have used these settings, and we have also used the "CallManager" defaults and made the neccessary changes to payload in the Nortel Gateways.

1. Downloading was not the correct term, we submitted our "upgrade" order and received the upgrade. We have installed the software on a non production server, and are preparing to test. We have also downloaded the spA.

2. I agree that the patches do not seem to apply.

3. Yes, I have attemped to use Fast Start to no avail, and I programmed the trunks as H.225. Inbound calls IP set to IP set work, but fail when connected to a resource such as a trunk.

4. Stay tuned............

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: