ietf
[Top] [All Lists]

Re: the VoIP Paradox

2003-09-02 17:18:55
> telephony
>
>      The public and private branch exchanges of the telephone network.
>      Availability in telephony is counted as /five nines/. So, 99.999% of
>      the time when you pick up the phone you get a dialtone. That's all
>      but five minutes per year.

and one has to carefully tune the measures of "availability" to achieve
that figure.

for instance (personal experience from Norway, mostly):
- pabx availability is dependent on the janitor not tripping over the power
cord, and not measured
- a cell phone that is out of range is not counted (of course!)
- a subscriber line that is down is not counted until the time the
subscriber reports the error.
- "acts of God" like fires in telephone exchanges (and even some "backhoe
fade" cable faults) are not counted.
- if you try to call Liberia, all bets are off.

in practice, five nines is a shared illusion between the operator and the
regulator.

It's an illusion in my experience as well. Growing up in Oklahoma, long
distance calls (that is, any call not to one of the 700 other people that lived
in the town of Perkins) could be direct dialed but then an operator had to come
on the line to ask for the caller's number (no, I'm not fooling about this). No
number gathered meant no call got completed. The failure rate for this setup
was, for whatever reason, non-negligable. I'd estimate its reliability at two
9s at most. The fact that the initial dial tone was always present didn't
translate into a reliable phone service, and worse, you could waste a
considerable amount of time before it was evident the call had failed.

Once I moved to Los Angeles I spent some time working as a technical director.
One of the things technical directors do is make phone calls -- lots and lots
and lots of phone calls. Unfortunately the PABX I was "behind" had an interop
problem with the telco with lots of finger pointing on either side as to  where
the fault was. The result was a 25% call success rate. The failure modes were
also many and varied, which added to the fun. (I used to keep a tally on a pad
next to the phone, so I had actual data to back up this figure.)

These days things are much better. But phone problems are nevertheless quite
common around here. On several occasions my DSL connection has worked fine but
there was no dial tone on the phone line. (In one case one of the conductors
was actually broken. The DSL circuit managed to work, albeit with a high error
rate, over a single piece of copper. POTS service did not. Ain't capacitive
coupling grand?) And in talking with other people who live around here, I find
that my experiences are hardly unique. If anything, I have had fewer problems
than most.

As far as power failures go, I have full battery backup at home, and when the
power goes my experience has been that both POTS and DSL have continued to
operate, but call completion rates tend to go way down.

Conclusion:

statistics always lie.

Either that or an awful lot of people must be experiencing truly spectacular
phone service to balances things out.

                                Ned

P.S. Nothing prevents you from having battery backup on cordless phones. That's
what I do, for the important ones at least.



<Prev in Thread] Current Thread [Next in Thread>