CCM Bandwidth Question

[JeffG1]

Here is the issue:

Site A is connected to the WAN with 256K of Realtime Bandwidth. Site B is connected to Site A with a t1 link. QOS is set for 768k on the T1.

Site A has a local PRI and DIDs.

Calls for Site B come into PRI and across the PRI.

Voice mail for both sites go out the WAN link at G729.

Site A Phones in one Device Pool, Region, Location Site B phones in another Device Pool, Region, Location.

Now in CCM I have site A set for 256K of Bandwidth, Site B set for 768k

Site B to Site A calls are G711.

When a call is made between site B and Site A, or out the PRI I know CCM subtracts from the 768K of Site Bs bandwidth. However does it do anything to site A's 256k bandwidth?

The reason I ask is because users are getting not enough bandwidth when calling between both sites.

Now this occurs late in the evening when there were few people in the office.

[Ayodeji Okanlawon]

Hi,

Callmanager determines if there are enough bandwidth nased ont he location bandwidth of the originating device...

Hence from site B to A, when 9 calls are in progress, the 10th call will receive not enoough bandwidth.

ALso from site A to B, a total of 3 calls can be made if u are using g.711 and 16 calls if you are using G.729..

Hence bandwidth deduction is from the originating device only.

[JeffG1]

Ok I thought that was the case, so I could increase the bandwidth settings for site A to match Site B 768K. This would then allow 9 calls from Site A to Site B. However this could cause an issue if 9 other people from site B then try to make a call the QOS policy would be over subscribed.

Also This would cause Site A & B to think it has 768K of bandwidth for Voice Mail calls. I could fix this by increasing the realtime bandwidth of Site A's wan link to 768K.

And of course I could set calls between site A and B to G729 and gain alot more calls...

Thanks

[Chris Deren]

Why are you not using G729?

Also, keep in mind that CM location calculation does not take L2 overhead into account, so assume you are using G711 for WAN calls, CM calculates each call to be 80Kbps, but with L2 overhead over PPP T1 it's more like 87kbps, so LLQ with 768kbps will allow for 768/87=8.82 calls, which you need to round down to 8 calls. Otherwaise if the WAN is congested and you allow 9 calls, QoS will kick in and all calls may suffer as random voice packets will not be prioritized.

HTH,

Chris

[Ayodeji Okanlawon]

Hi,

Chris just nailed the final thing.L2 overhead! You need to add layer 2 overhead as chris has mentioned and make sure that your LLQ must only allow 8 calls to avoid packets been dropped.

The best way to do this is this..

With L2 overhead, only a total of 8 calls will comform to your LLQ bandwidth, hence you should also providion callmanager to allow only 8 calls.

So your location bandwidth should be eithr 80X8 or 16X8 depending on what codec you detrmine to use.

So this way you have forced callmanger to conform to your LLQ bandwidth int hat it will only allow 8 calls and this way your LLQ will not police and drop packets.

HTH, pls rate useful posts