You can also try calculating bandwidth at www.erlang.com
John Deatherage wrote:
> Some things that might help: 1. Are you using CODEC like G.729 that
> works at an 8Kbps bit rate? If not, you'll be hating life.2.
> Bandwidth at full rate & Multilink PPP (MLPPP) or FRF.12 will still be
> 17.2kbps.3. You need QoS, even if you can burst to port speed at
> 56kbps.4. Consider LLQ for Frame Relay for your QoS, which is now
> supported in IOS as of 12.1(2)T. We had to use this to get good voice
> quality on a moderately saturated FR link with a 1.544mbps CIR.5. You
> may need QoS on your LAN if the links are congested near the 3640.
> Look at QoS as an end-to-end solution.6. Perform ping tests from the
> 3640 to your remote sites to get a better idea of how many ms the
> round trip is. Try pings from both sides. Things to keep in mind:-
> VoIP overhead will kill you - you need to do testing to figure out
> exactly how much bandwidth you are using.- Good luck doing this over
> 56kbps, even if you use the full line.
>
> ----- Original Message -----
> From: Chris Boyd
> To: [EMAIL PROTECTED]
> Sent: Friday, November 10, 2000 1:08 PM
> Subject: VOIP Troubles
> All right guys I need some help.... I have been working
> with Cisco for a while now on a VOIP issue. The problems
> lies in both call disconnects and voice distortion. We have
> followed all the steps for traffic shaping (QOS) and rtp
> header-compression but these do not seem to help. We have
> 150 remote sites all running 2600's with FXS modules that
> all come back into the host site where we have 2 7206's.
> Each of these links are 56k frame-relay links with 16k CIR
> running very few applications mostly small transactions and
> Citrix clients. The call must then traverse two internal
> Ethernet segments, routing through our 6509 backbone switch
> and then into a 3640 before hitting the PBX. Cisco seems to
> think that we need to increase our bandwidth to support the
> voice traffic, however, that is not something I have been
> able to sell to the "powers that be". We sold this idea on
> cutting cost and in our estimations for upping the CIR to
> even 32k will be significant cost increase. Right now I am
> shaping to 16k with an 8k committed burst so at any one time
> I should be able to burst to 24k. Assuming that I am able
> to burst to port speed (56k) why would I have call
> distortion unless there is some latency coming through the
> ISP's switch? We also have another company site that also
> comes back in this way and we have no problems with those
> calls.O.K. that being said (and hopefully not too confusing
> to follow) here come the questions: 1. Has anyone else
> implemented VOIP in slow links successfully?2. Is anyone
> else having QOS problems with their VOIP
> implementations?3. Do I need to prioritize the voice
> traffic through the local network?4. Has anyone tried
> turning off traffic shaping and letting the voice and data
> compete for bandwidth? Thanks in advance for your
> feedback!
> Thanks, Chris Boyd, CCNA
> Network Support
> Alex Lee, Inc.
> 120 4th Street SW
> Hickory, NC 28601
> (828) 323-4103
> http://www.alexlee.com
>
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]