[e2e] Re: crippled Internet
vjs at calcite.rhyolite.com
Thu Apr 26 06:36:39 PDT 2001
> From: Randy Bush <randy at psg.com>
>>] According to G.114, one-way delays up to 150 ms should cause no problems
>>] with typical voice communications. From 150 to 400, loss/noise/etc. become
>>] a bigger issue and "long-delay" echo cancelers are needed. (But some form
>>] of echo canceler/suppression is needed on ALL delays beyond 25 ms.)
>i admit i am blind and stupid, but i just can't wrap my head around how this
>applies to voip. echo cancellation?
>>Given the sort of digital processing against echos that is now routinely
>>done for good speakerphones and the CPU cycles available in modern
>>personal computers, I wonder if the old worries about echo cancelers are
>>irrelevant. There's also the fact that everything is digital in VoIP.
>it seems more relevant to me that voip is two *separated* one way paths.
>what's to echo?
I did say the the token ring/100VG line was a calumny, didn't I?
But it does have a grain of truth. What if the VoIP path involves
old fashioned, 2-wire, single pair analog telephones at either end?
For example, what if you call someone through a VoIP-to-telephone
gateway? What if the sound bits never go near a telephone, but either
or both users typical computer speakers and microphones instead of
head sets? Then you get echos.
In real life, most VoIP calls will involve one and usually two old
fashioned 2-wire analog telephones with old fashioned echo cancelers
because of economics. My experience is that the freeware VoIP works ok,
except when one retail ISP or the other or an interchange point to
introduce decides to introduce many 100's of ms of delay or high packet
losses, or one PC box or the other decides to run something else. Given
the cost of phone calls today, why bother with the hassles of setting up
a special VoIP call between computers?
Telephone time is now too cheap to meter unless you are running a big
company's system and deal with 1000's of hours/month or you buy the
wrong wireless plan. That's proven by the U.S. television advertising
campaigns for stupid plans for "$1.00 for any call up to 20 minutes
and $0.07/minute after that." Any consumer who cares is paying
$0.05/minute or less all for all (non-wireless) time (in the U.S.).
If people won't bother to shop for cheaper long distance time, they're
not going to both being tied to their PC's.
Another problem is that the standard junk PC system software and access
routers don't do the obvious and easy QoS queuing that reasonable UNIX
PPP and SLIP code has been doing for 12 or 13 years. Windows or Windows
applications also can't figure out how to keep high priority processes
running. If you and the other party are using free VoIP PC software
and one of you has an analog modem or runs Windows and accidentally asks
their computer to do something else, your "connection" gets very bad.
Vernon Schryver vjs at rhyolite.com
More information about the end2end-interest