Document ID: 13903 |
Introduction
This document addresses frequently asked questions related to the Cisco MC3810.
Q. Do I need an MFT (Multi-Flex Trunk) module to enable Voice features on the MC3810 multiservice concentrator?
A. No, not unless you are planning to run Voice over ATM (VoATM). ATM encapsulation is supported only on the MFT module.
Q. Can I configure a portion of the MFT as digital voice ports?
A. Yes, but with two caveats:
The total number of simultaneous calls cannot exceed the total number of calls permitted by Digital Signal Processors (DSPs). DSPs reside on the Voice Compression Module (VCM). Each VCM has 6 DSPs. Each DSP can support two G.729a calls, or one fax call. There is one VCM in the analog configuration and two VCMs in the digital configuration. Thus, the maximum number of simultaneous G.729a calls is 12 or 24, for the analog or digital configuration respectively.
An analog personality module or a digital voice module must exist before voice commands are visible in the parser.
Q. What are the rules for configuring the data segmentation size over Frame Relay for voice transmission?
A. The number set for the data segmentation size command is the single most common factor in voice performance on the MC3810. Poor performance can result if this number is either too small or too large, which is why there is no default size for this command. If the fragmentation size is set too large, the voice will sound "choppy" because voice samples must wait behind large data packets during egress. In addition, the voice will sound fine until you add data. If the fragmentation is set too small, the voice will probably sound fine, but the MC3810 will be over-burdened with fragmenting packets. This becomes a problem on higher speed links.
For example, a fragmentation size of 80 is fine on a 64 k link. However, if you use the same size on a T1 or E1, the voice will sound great, but the router will be severely taxed fragmenting frames unnecessarily. The result will be less than T1 throughput. The correct fragmentation size for a T1 is 1600 bytes. The rule of thumb is to set the fragmentation size to slightly more than the link speed. Links over 768 k do not usually need fragmentation. See the Recommended Data Segmentation Sizes in the MC3810 User Guide for link speeds and fragmentation sizes.
The formula to calculate this number is:
Data_Seg_size = ((Cir_Rate / 8) * Jitter_Delay) / 3Where:
Data_Seg_Size is the segmentation size in units of bytes.
Cir_Rate is the pvc cir rate in units of bits per mili-second (bms).
Jitter_Delay is the maximum queuing delay per node.
3 is the output queue depth for the device controller transmitter.
For all calculations, it is assumed that jitter delay is no more than 30 ms per node.
Q. What could be the reason behind getting "PEER out of service" message?
A. In MC3810 with an IOS prior to 1207XK1, you need to match the codec types in vofr and pots legs. The default codec used in pots and vofr legs is G.729ar8 and if you configure a non-default codec for the vofr dial-peer, you need to configure the same codec under the voice-port as well. Otherwise this error could be seen in the debug messages. This is MC3810 specific and you do not get this anywhere else.
Related Information
- Voice Technology Support
- Voice and Unified Communications Product Support
- Troubleshooting Cisco IP Telephony
- Technical Support - Cisco Systems
| Updated: Feb 02, 2006 | Document ID: 13903 |