dslreports logo
 
    All Forums Hot Topics Gallery
spc
Search similar:


uniqs
843
Warez_Zealot
join:2006-04-19
Vancouver

Warez_Zealot

Member

[Info] Cisco Video Conference Bridge Quality Issues

I am working on finding a provider that has a backup bridge/VCS hardware we can rent should our primary hardware (local in Vancouver) go down/fail.

In the end we decided to use a provider in Toronto. All the clients/codecs that will connect are located in B.C, so the traffic will go from BC > TO > BC.

Doing our tests we do notice poor quality when there is fast movement on screen, or playing media.

The latency is approximately 50-60ms over an MPLS circuit (traversing our providers backbone *not over regular internet*)

We are using an HD and SD bridge, and the quality is bad on both (HD more so). I don't see any packet loss, but I think the issue is a mixture of Jitter sourcing from the bridge when it sends all the VC packets to the B.C codecs.

I was using Ping Plotter on a terminal server located on the VC network where our test codecs were with a session going to the HD and SD bridge. I did see Jitter jump during the testing, but I don't want to say it's 100% that until I can get more info.

I'm kinda new VC related services, but it seems to me the main problem is the bridge being geographically so far away from B.C where the actual codecs are located.

Does anyone have advice/testing tips so I can prove/disprove this theory?
HELLFIRE
MVM
join:2009-11-25

HELLFIRE

MVM

Ping is the most basic testing you can do, so definately use that. If possible, set up a station on the same network
segment as your bridges to run the ping tests / ping plotter.

Don't know if the bridges themselves are amenable to being polled by SNMP; if so, it'd be worthwhile to record their
response time, just as another thought.

Don't know if you have any IOS devices in your environment, but running IP SLA is another option to record jitter...

Beyond that, it's proving your gear out, and working with your provider in proving their side out as well. To give
them something to chew on, DEFINATELY record exact date / timestamps where latency / jitter is noted so they can
backtrack.

My 00000010bits

Regards

TomS_
Git-r-done
MVM
join:2002-07-19
London, UK

TomS_ to Warez_Zealot

MVM

to Warez_Zealot
It would be worth making sure that the switch ports the VC units are plugged in to are running at full duplex. I used to deal with a lot of similar issues in a former job life and a lot of them came down to half duplex Ethernet, in particular where the switch side had been hard coded full duplex which then disables autoneg causing the VC unit to go half duplex.