How relable does the Internet need to be? (Was: Re: Converged Network Threat)

Sat Feb 28 13:54:00 UTC 2004

At 03:21 AM 2/28/2004, Adam Rothschild wrote:
>On 2004-02-27-18:43:50, Daniel Senie <dts at senie.com> wrote:
>[...]
> > >With Vonage you have to tell them where you are located so they can set
> > >your 911 service up to the proper 911 center.
> >
> > You can take your Vonage with you. Some people do this. It's a bad idea to
> > dial 911 on a Vonage setup that you have on the road with you, but the
> > question of mobility certainly can be studied using this service.
>[...]
>
>I'd take your argument a step further, and argue that it's a bad idea
>to blindly rely on any carrier, VoIP or otherwise, to provide 911
>services, period.  Test it periodically (and of course, make your
>intentions clear, and confirm that there is no real emergency to
>report), lest you'll find yourself in a position where you need to use
>it, but can't.

Testing 911 or equivalent on a periodic basis is actually a good way to get 
in trouble with local authorities. In at least some jurisdictions (the 
community I live in, for one), dialing 911 gets you a visit from the police 
no matter what. They are required to roll on all 911 calls.

>Vonage's E911 implementation (which in actuality is outsourced to some
>provider whose name I forget offhand) is broken in too many ways to
>enumerate in this forum without totally losing it and committing
>multiple NANOG AUP infractions.  Try it; chances are you'll be routed
>to your local coast guard platoon commander, or the highway patrol, or
>dead air, or just about anywhere other than your loca PSAP.  Not to
>merely single out Vonage, I've had my fair share of encounters with
>other PRI- and VoIP-based CLEC's who are equally broken, or at best,
>will hand off your calls to the proper PSAP, but with blanks for all
>informational fields except your BTN.
>
>Our 911 system is broken and vulnerable enough as-is.  I question the
>business ethics of anyone who would want to add additional layers of
>complexity and inadequately tested technology (like... *cough* VoIP),
>risking life and limb in the process, and say it will be there for you
>when you need it.  And, much as it pains me to say it, things aren't
>going to get better until the FCC comes down hard, and fines all the
>basement establishments with 2 BRI's and a mic^H^H^HAsterisk box who
>purport to offer [E]911 service and in fact... don't.
>
>Finally, quick show of hands for you RBOC's and cable operators out
>there... how many of you can say with confidence you've got battery
>backup on all your DSL-enabled RT/DLC's?  Cable headends?  Active HFC
>distribution gear and amplifiers in the field?  I rest my case.

It's just as bad on the wireline side. We've got a number of SLC-96 mux 
units on the POTS network here. The batteries in the units have gone bad, 
and not been replaced. When the power fails, the phones go dead.

There's a dual-position knife switch under the pole-mounted SLC units here. 
The lower position is to connect to a portable generator which in theory 
the telco would position in the event of an extended power failure. However 
after an automobile accident that damaged one of these, the replacement 
cutoff switch no longer has the generator port. Coupled with the dead 
batteries, there's really nothing worthwhile in the way of power outage 
handling in the POTS system, any more than there is in other technologies.

In recent power failures, the cable modems continued to work, but the 
Verizon phones did not. So, we had our Vonage line, but not the POTS lines. 
(We have an auto-start generator in addition to UPSs with good batteries, 
so we had power ourselves).

Cell sites have backup power, but in many cases no generator. When the 
batteries go, the cell site drops out.

The question in all cases is what is the level of service acceptable to 
regulators and emergency services coordinators? Clearly there are problems 
of both power and call routing which must be addressed. It's unlikely NANOG 
is the forum for specifying standards in this area. It is similarly 
unlikely the IETF is the appropriate body, though it may be a place to 
figure out how to meet the requirements specifications of some other body.

> > The simple response to this situation is to say "tough luck. Use a
> > wireline phone."
>
>Yes.  Very much agreed.

You took that out of context. I was not making the argument that such a 
statement was sufficient. For those developing VOIP technologies, they are 
going to have to come up with a better answer. As you note, the 911 system 
has a lot of problems today. As I noted, the wireline network has the same 
power vulnerabilities as the VOIP and cable-based telephony services and 
wireless services.