On March 30, 2005 05:24 am, Obihuan wrote:> My calls, depending the hour of the day, have diferent quality.
> Sometimes I felt cuts in the conversation or lost the sound on one of
> the end point.
>
> All of the providers I tested had any kind of trouble.
Sounds like the trouble is on your end then. I use nufone almost exclusively
and put about 5000 minutes a month through them, with multiple simultaneous
calls (mid-size business) and while I occassionally have some audio problems,
I have never had issue with nufone's network. I have been able to (in my
mind anyway) prove that the connectivity issue was on my end, as when the
problem occurs it occurs with any provider I happen to be using, and they all
take wildly different paths once it leaves my (decently connected) internet
provider.
> My internet gateway is an 1 Mb. ADSL conection y I make QOS by the
> router 70% of bandwidth for SIP and IAX2 protocols and 30% for others
> protocols. With 3 simultaneus calls.
>
> I thing that the problem is in the providers side, cause we make calls
> between our
> diferents offices via IAX2 without quality problems, but I am not sure.
> I said that because when in US the people wake up and start to work,
> about local time 13:00, our calls get more troubles, like cuts, but
> before that time our calls goes better than after.
Is there any heavy downloading or uploading going on around that time? The
unix program 'rate' or even tcpdump or ethereal should be able ot help
you
determine this. Remember that you can only rate-limit your OUTGOING traffic.
Traffic headed for you can be dropped in an attempt for tcp's automatic
backoff to slow down the connection, but as the name implies it only works
for TCP.
Feel free to try my traffic control script:
http://www.mixdown.ca/~andrew/dump/rc.tc -- it runs on our upstream router
and with it I am able to keep our connection loaded but still have voice
traffic pass through as top priority. Again, it tries to limit the incoming
traffic but that's more based on luck than anything else. :-)
-A.