I should probably also point out that the public internet links also
tend to be far larger than the private interconnections and thus
infinitely easier to scale than needing to continually flip production
voice traffic onto ever larger private circuits.
In this case, the network to which Miles is referring, has gigE to five
providers and one public peering point and thus has sub 1ms latency to
the networks of many of the major origination and termination players
(via multiple paths) as such could easily be construed as better
connected than a private link.
What does astonish me is what people will do with their production voip
traffic in terms of running it across either small links with no QOS or
providers known for peering issues. I can tell you I would never run
production voice over a consumer circuit like verizon Fios or random T1
link and expect flawless results. I use voip in production at my home
office and even there I have provider-side QOS enabled on the link to my
house, without control of the link the connection is coming in on there
is no guarantee of a congestion-free (or QOS enabled) path which is
required for great voice quality.
There also appears to be a bit of a misconception of the "downsides" of
running this kind of traffic over the public internet. As miles noted
originally, the congestion very rarely happens at the core, it happens
at the edges. This is not just a "guess" this is fact, I have tracked
latency towards big destinations from our network for quite some time
and I can tell you that the peak-to-lull latency change for most large
destinations is around 150-200 microseconds (yes, 0.15 to 0.2 ms) if you
are doing the tests from a well connected place (the above described
network in this case) whereas many consumer-type connections have large
latency variations depending on load. The only time I have ever seen
congestion in the core was across peers between large tier1 providers
and in the last three or four years I have only seen that on peers to
AS7018 (ATT) or AS174 (Cogent) which at least in our case accounts for
virtually none of our voice traffic.
Anyway, that is it for my contribution. As miles notes, we offer
wholesale origination services using our own IP backbone with SIP
proxies located in the Westin Building in Seattle, WA. Feel free to
try traceroutes or speed tests.
http://as11404.net or
http://as11404.net/speedtest (be nice to www.as11404.net as it is only
on 100 mbit/sec Ethernet).
It is getting late, so I hope all of that made sense.
John van Oppen
Spectrum Networks LLC
-----Original Message-----
From: asterisk-biz-bounces@lists.digium.com
[mailto:asterisk-biz-bounces@lists.digium.com] On Behalf Of Miles
Scruggs
Sent: Tuesday, June 24, 2008 10:31 AM
To: Commercial and Business-Oriented Asterisk Discussion
Subject: Re: [asterisk-biz] VoIP Provider
You can't really change the way a provider offers their circuits, no
matter how much you want to speak a certain way for personal
preferences. They offered it over T1s or DS3s, you can try hammering
out semantics with them, but I doubt it will help.
On Jun 24, 2008, at 10:14 AM, Steve Totaro wrote:
> It would seem if someone is speaking of T1s then they should also use
> T3, if they are speaking of DS1 then DS3 would be the proper naming
> convention.
_______________________________________________
--Bandwidth and Colocation Provided by http://www.api-digital.com--
AstriCon 2008 - September 22 - 25 Phoenix, Arizona
Register Now: http://www.astricon.net
asterisk-biz mailing list
To UNSUBSCRIBE or update options visit:
http://lists.digium.com/mailman/listinfo/asterisk-biz
_______________________________________________
--Bandwidth and Colocation Provided by http://www.api-digital.com--
AstriCon 2008 - September 22 - 25 Phoenix, Arizona
Register Now: http://www.astricon.net
asterisk-biz mailing list
To UNSUBSCRIBE or update options visit:
http://lists.digium.com/mailman/listinfo/asterisk-biz
Posts
