On Thu, 21 Jul 2005, Elizabeth Zwicky wrote:

> On Jul 21, 2005, at 1:55 PM, Michael H wrote:
> > Well here's the deal, they will also route our data traffic so we will
> > be eventually switching off our current T-1. We are also getting two
> > T-1's for some redundancy since I'm a little nervous.

> How much experience do you have with redundant T-1s and failures?

/me jumps in


> In general, if you just buy 2 T-1s (even from different vendors, unless
> you pick the vendors VERY carefully), they will be routed across
> the exact same pieces of hardware for most of their lengths. The backhoe
> that goes through one will go through both. The switching hardware

Different wireline carriers are only easy to get in some locales, and are
certainly one of the things folks should look at. Post 9/11 it's getting
more and more difficult to get a look at the fiber maps for an area,
which makes carrier selection more difficult.

For new buildings, I try to do different ingress/egress points too.

> failure
> that kills one may well kill both. It's not uncommon for

I do have to say that I haven't seen that many switch failures in recent
years. It always seems to be a backhoe in an urban area or a span failure
on a pole in the middle of the circuit.

I spent a lot of the last month dealing with red alarms on data and voice
T-1s because the LEC wasn't as interested in fixing things as the CLEC the
customer bought the lines from.

A new data line got sent back to the splicers *twice* (it's great
when the LEC's tech really does want to do it right![1]) before the
circuit was sent to the CLEC. While the CLEC tech was in for acceptance
testing, he noticed the voice Ts doing sporadic alarms, and we had a
redundant data T set go out. For most of last week though, only one of
the data Ts was solid red. That's a first for me, as I usually don't
advise redundant circuits from the same carrier (but I didn't set this
place up.) Not that it's an endorsement for redundant Ts from the same
carrier- I'd still do the ILEC and a competing LEC (and there is one here,
though there wasn't when the circuits went in) if it were up to me.

After a month of escallating, griping and complaining, the LEC seems to
have gotten its act together and fixed the lines.

We've generated one PUC complaint, and about 12 tickets in a 5 month
period, but things are finally right. Of course, late last week there
were issues at the other end of one of the circuits. *sigh*

> 2 different vendors to be re-selling strands of the same cable owned
> by some third party. Even if they are in different cables, they will
> usually be in the same conduit.>
> The really interesting question, however, is "What are the odds your
> phone
> line is in that conduit, too?" Because if that answer is above 75%,
> maybe
> it doesn't matter that your redundant T-1s aren't redundant. In my
> experience,
> redundant T-1s fail together more often than T-1s and phone service, but

I find that a little odd, since phone service is normally just another T-1
from the customer to the same CO. Though I do find that data T-1s fail
more often than voice, generally that's because the voice side recovers
more easily on poor quality lines (that is both fail, but only the data
failure is noticable.) Even small phone switches seem to do pretty well
getting around transient line failures that error data out completely.

> probably not even by a factor of 2, if you buy the T-1s from different
> vendors
> so that the bozos dealing with the software don't cut you off. If you
> buy
> the T-1s from the same vendor, there's no interesting redundancy at all;
> they almost always fail together.

If you're in a major city and you're not doing something that'll go round
a SONET ring, that backhoe's going to get you sooner or later. VoIP may
give you something like FX service, but I think you really have to think,
plan and pay for anything that's going to have serious redundancy.

[1] The circuit tested "within spec," so he could have just left and
called it done the first time.
Paul D. Robertson "My statements in this message are personal opinions
paul@compuwar.net which may have no basis whatsoever in fact."

firewall-wizards mailing list