[asterisk-users] Call Quality Issues With 2 Trixbox's - Router Issue?

[Daniel Cole]

Hello Everyone,

We have recently installed a pair of Trixbox servers in for a client of our. They have two locations, with one server each. The servers terminate 3 standard POTS lines into a Sangoma A200D card. The servers are IBM x3250 1RU servers (1GB Ram, Raid 1 160GB HDD, Dual Core Xenon Processors). We are using Trixbox 2.2, and G729 all around.

Each site has two (2) 512k/512k ADSL connections terminating into a Cisco 877W router (using an additional 'dumb' modem in a separate VLAN for the extra dsl connection). Using policy based routing, all Voice Data goes over one DSL connection (the one that terminates directly into the router), and all other traffic (e.g. Web and VPN) goes out the second connection (the bridged dumb dsl modem).

We are also the ISP for this client, and as thus we have full monitoring of our Layer 2 and Layer 3 networks. From our analysis, it doesn't appear that there is any issue in these networks. We have other customers using the VoIP service, who have not complained of these issues.

Now for the Fun part!
The client is complaining of issues with inter-site calls. They are reporting issues with crackly and broken speech, and horrible jitter (or packet loss). This presents a huge issues, because they have one receptionist answering all calls for both sites. So if a call comes in from the other site, it automatically an inter-site call, and the quality falls out of it. If the call is then transfered back to the originating site, the audio 'bounces' between the two sites, which add to the call quality degradation.

We have been monitoring the router while these incidents have been reported, and it does not appear to be a bandwidth issue. The DSL tail used for Voice gets to no more then 120k in each direction (we have tested the links, and can pull data at 53k/s between sites). CPU usage floats at around 20-25% under load. The router has only shows major packet loss (that we can tell) when REALLY pushing it in testing (e.g. 10+ calls between sites).
We have enabled the SIP jitter buffer, as well as the IAX jitter buffer, which appeared to make a huge difference, but the issue is still ongoing.

These issues have also been reported with some outbound VoIP calls. Internal calls, and calls directly in or out of the Sangoma card are clear, with no issues reported.

Does anyone have any thoughts on what could be causing these issues? We have been racking our brains here, and have tried everything that we can think of. These system is a million times better then what is what when it was first installed, but it is still not where it should be in terms of quality.

Any thoughts/ideas are most welcome.

Thank you


Daniel Cole  (CCNA)

 P Please consider the environment before you print this e-mail or any attachments.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.digium.com/pipermail/asterisk-users/attachments/20071212/b2923ecf/attachment.htm