pwlanman

My company has 2 sites, one in Chicago and one in Minneapolis that are connected with a private line T1. There are currently 8 DCS channels for voice (4-digit dialing), and the remaining half of the T1 is for our data network side. This setup has been working well for the last 2 years. We have been working with our VAR to switch from the DCS channels to straight VOIP, so we can take advantage of unused voice traffic for data. When we made the switch, it seems to work fine for a day or two, then calls start to suddenly drop and voice quality degrades. We've switched back and forth between VOIP and DCS after each software and firmware upgrade and the problem still persists. We have been monitoring the traffic with a network sniffer program when VOIP is running and through the whole time, we get ICMP errors: destination unreachable, port unreachable – from the Minneapolis MedPro card. We had the network tested and the link is good. Also the outages do not seem to happen if there’s a lot of data traffic. I would almost expect this kind of performance on a VPN or frame-relay, but not on a private line T1.

Here’s some more info:
Both site’s MedPro cards are TN2302AP, HW03, FW071.
Chicago’s C-LAN is TN799DP, HW00, FW009.
Minneapolis’s C-LAN is TN799C, Vintage 3.
The network switch ports are set at 100 full for the Medpros, 100 full for Chicago’s C-LAN and auto for Minneapolis’s C-LAN.
The routers are setup so the IP addresses for both Medpros & C-LANs have priority over everything else (data). The voice traffic is using the compressed codec.

So far Avaya support and our VAR have not been very helpful. The VOIP side is still kind of new to me, so any ideas or companies that know VOIP and Definity switches would be appreciated.

mallan

Just a few quick questions:
1) Do you have a dedicated VLAN for voice traffic, or is it riding on the same IP addressing scheme as data?
2) You mentioned compressed codec - what are you using? G.729?
3) What Quality of Service features have you turned on in your data switches/routers, and what type of equipment is that?

PBXtech.info

 

bharris

The phone switch needs to be running rel 10.5 or higher or the autodetect feature of the 10Mb CLAN and MEDPRO will try to reset the network switch or router port speed so that they are running at the same speed. From what I understand Avaya has set these at 10Mb fixed for the CLAN cards even if they are 100Mb cards. It's more inportant for the MEDPRO card to be running at 100 Mb but both sides need to be the same speed. Use the "Change Ethernet" command on both of your phone switches to adjust these settings.

pwlanman

That may be it. The Chicago switch is running rel 10.5, but Minneapolis is on rel 9. I can't even adjust the speed on that switch. I'll test with another site that's running rel 10.5 or 11.
Thanks.

pwlanman

1) It is using the same IP address scheme as data.
2) Yes - G.729
3) Highest priorty queuing on the routers & no qos on the switches. The routers are Cisco 3600 series and the switches are HP Procurve 10/100.

bharris

Also noticed that your Chicago CLAN card is running FW 9 and that release has some problems so you should upgrade to the new FW 10 release. I have put it on some of our systems last week and it seems to be running fine. Release 10 is the first version to be fully compatable with Communications Manager 2.0 It looks as if you are rumming the latest release of firmware on the MEDPRO cards.

The CLAN at Minneapolis is a fixed 10Mbps card only so with system release 9.1 you will also have to put that systems MEDPRO at 10Mbps by locking it at the network switch or router port. Also the Minneapolis CLAN is a hardware rel 3 and the current release is ver. 6 so there could be some issues there also. We use a few of the same cards but we try to only use them at sites that only need administrative login over the network and not for heavy VOIP traffic.

PBXtech.info

 

pwlanman

So does the network switch for the MEDPRO need to be set to 10Mbps because it communicates with CLAN via ethernet, even though they're both in the same PBX?

stubborne

They connect to the network independently from each other. The Medpro can run at 100/full while the CLAN is 10/half. It's been my experience that the Medpro can revert to auto/auto if a reset or some other interrupt happens. If your switch is hardcoded to 100/full, you will have issues there - perhaps dropped calls or dead air. I'd check the switch ports for errors to see if there is a mismatch.

__________________
Stu

bharris

If the two cards are running in a system with software rel less than 10.5 then both cards will need to be set to 10Mbs or else the network switch or router ports will keep getting request to re-negotiate the speed. With software rel less than 10.5 you can't set the speed of the cards in the phone switch so you have to lock down the network ports. Some network switches and or routers may handle this better than others but we have had this happen with Cisco and 3-Com until we upgraded to software rel 11.1 last year.

pwlanman

We configured VOIP from the Chicago switch to another location (L.A.) that had rel. 11. It's been up and running for almost 2 weeks now with no problems. We're going to upgrade the software to rel. 10.5 on our Minneapolis switch. Release 9 seems to be the problem.