Anyone using 30ms?

[Stinney]

Anyone out there using G.729/30ms as their primary codec?

Have a situation where we're being asked to change our entire environment to 30ms.

Any gotchas from anyone who's had to or tried to do this?

- Stinney

“The man who asks a question is a fool for a minute, the man who does not ask is a fool for life.” - Confucius

 

[newglide]

I tinkered with 729AB going to one of my remotes and I had issues with audio between the host and the remote.
Looked up some tech docs and it appeared to be a bandwidth setting.
I was using it to trouble shoot another issue so I didn't stay on it very long.

You can always test it out by configuring another network region and only adding a couple of phones to that region.

 

[wpetilli]

I used this codec for a remote office that didn't have alot of WAN bandwidth and have had no issues. As long as you can apply QOS it isn't really that much different to the ear.

 

[Stinney]

How about for adjunct systems that are connected like Voicemail, Call Recording and IVR.

- Stinney

“The man who asks a question is a fool for a minute, the man who does not ask is a fool for life.” - Confucius

 

[kyle555]

Well, it's only AAM 6.3 that supports 729 at all. For your own sanity, i'd reckon that those adjuncts be behind gateways and intervening network regions that make sure they only get 711.

Depending on how/what your recorder and IVR uses, there's sure to be considerations. I'd try to leave that to G711 if possible and just be careful about how many transcodings you do back and forth as that can affect quality.

 

[jimbojimbo]

Yes, there are several potential issues.

Many carriers do not support 30ms (or only support it across some types of calls). This means you would not be able to shuffle the call since you need both 20ms and 30ms so there goes your DSP utilization.

Why are you looking to change your "entire environment". Doesn't make sense. If you have a specific device requiring 30ms setup a codec to use 30ms only when communicating to the specific device.

The only reason I can see you being asked to do this is if the routers are being overloaded by the large number of small packets. Most stated router capacity does not use small packets to make the determination. You can expect up to 70% less than stated capacity when using just for voice.

The only time I typically use >20ms sampling rate is for Call Recording (which is going to anchor anyway) and set it for 60ms which is then generates only 1/3 the packets.

 

[Stinney]

jimbojimbo,

"Smarter" (telecom) heads prevailed, but originally we were being asked to change our entire environment to 30ms because that's how AT&T "preferred" to deliver the calls in our new SIP environment to maximize the bandwidth utilization of our current circuit. However, because of the scope of the work to do this and the fact that we will be expanding that circuit's bandwidth 10x in the very near future, we showed that there is no need to do this change.

- Stinney

“The man who asks a question is a fool for a minute, the man who does not ask is a fool for life.” - Confucius