nanog mailing list archives

Re: SIP trunking providers


From: Rafael Possamai <rafael () gav ufsc br>
Date: Mon, 20 Jul 2015 10:21:28 -0500

When I originally posted the thread, I had asked Chicago due to physical
proximity, and my assumption being the lesser the number of hops, the lower
the probability of running into issues (latency, jitter and congestion). On
the other hand, one of my sandboxes are out of Las Vegas and I haven't had
any issues yet, but the number of test calls I've ran aren't enough to say
with confidence that distance and hops don't matter (indirect ways of
measuring latency, etc).

Another thing is, having your packets stay in Chicago and in Chicago only
is a nice thing, the efficiency of your overall system would be higher for
what it's worth, but as an example, the 2nd hop this e-mail is taking to
get delivered to Nanog is about 100 miles, who knows about the other ones.



On Mon, Jul 20, 2015 at 8:49 AM, Naslund, Steve <SNaslund () medline com>
wrote:

End to end delay is not the most limiting factor.  Jitter is the issue and
packet drops are the other issue that matters (more importantly the
distribution of drops).  I think the best reason to select the local
provider over the distant one is that the sooner he gets off the IP network
the less impairments he will run into.  The TDM network as antiquated as it
is, is less susceptible to congestion and call impairments than an IP
backbone network is.  I can tell you from running a bunch of International
VOIP networks that they are just not as reliable as TDM.  The average
internet connection just does not meet the reliability standards that the
TDM voice network has achieved.  IP networks are affected by congestion and
routing issues whereas the TDM network seldom has these type of problems.
An outage on a TDM circuit rarely affects other TDM circuits so they see a
lot less higher level outages.  I can understand why he does not want to
haul his voice cross country over IP when he is exiting locally most of the
time.

Yes, I understand that the carrier might very well be hauling that traffic
via IP even after he gets to his gateway point but at that point it becomes
their problem to deal with.

Steven Naslund
Chicago IL


If you’re going to the PSTN, who gives a shit where you do the
interconnect as long as its within 100ms.

If most of your calls are VOIP<->VOIP within Chicago, then it makes some
sense to set up a box and just send the external calls out to the trunking
provider where >you no longer really care where they are.

Absent significant network  suckage, there’s no place in the contiguous
US that isn’t within 100 ms of any other place in the contiguous US these
days.

Owen




Current thread: