Discussion point: Telco quality?

This latest fiber cut just kind of jarred my brain into the
realization of what "telco quality" services really boils down to.

In the end, all services, both traditional telephony services and
data services, are more alike than people give them credit for. :slight_smile:

- paul

That is correct. Especially if you live in New York City. Our success
rate lately on DS1 installs is around 30%. That's truly pathetic for the
price paid. Add to that about a full day of calling various numbers of
various CO clowns, and your install price has doubled.

Now I've never had a problem with getting a $15/mo POTS line working on
the first try... But then again it's not <sarcasm> "new" </sarcasm>
technology like them thar DS1 lines.

Charles

Date: Thu, 08 Oct 1998 19:21:19 -0400
From: Paul Ferguson <ferguson@cisco.com>
To: nanog@merit.edu
Subject: Discussion point: Telco quality?

This latest fiber cut just kind of jarred my brain into the
realization of what "telco quality" services really boils down to.

In the end, all services, both traditional telephony services and
data services, are more alike than people give them credit for. :slight_smile:

- paul

=-----------------= =

Charles Sprickman Internet Channel |
INCH System Administration Team (212)243-5200 |
spork@inch.com access@inch.com |

= =----------------=

And there's the rub. It isn't like DS1's (or anything else for that matter)
is any more "new" than they have ever been.

- paul

When I first went from MCI to PacBell/ACN an of PacBell veteran picked up
the hand-set and made me listen to it. He then asked me what I heard, I
said "dial tone", he said, "No, what you heard was a freaking miracle!". At
the time, our job was to make the new broadband HFC network as reliable,
but cheaper, than the local-loop system. As it turned out, he was right.
Building such a system, and keeping it running, decade after decade, *is* a
freaking miracle! We had 10,000 customers before SBC pulled the plug.

Paul:

Although the telcos talk a great deal about SONET rings and how reliable the
telco circuits are and so forth its is surprising how few major Internet
trunks are carried on SONET rings. And yet the Internet manages to carry
on. Despite some local inconviences mot Internet users never notice these
outages which goes to show how resilient the Internet really is.

It begs the question then, if the Internet can work quite well today without
all the supposed telco 99.99999+, do we really need telco "reliability" -
SONET rings etc?

As you know it is our contention that you don't need it. We believe that a
99.9999+ reliable Internet can be built without the underlying (supposedly
"relaible") telco infrastructure.

Bill

Although the telcos talk a great deal about SONET rings and how reliable the
telco circuits are and so forth its is surprising how few major Internet
trunks are carried on SONET rings. And yet the Internet manages to carry
on. Despite some local inconviences mot Internet users never notice these
outages which goes to show how resilient the Internet really is.

Funny you should mention this. Actually, aren't most data circuits (excluding
copper pairs into the home) fed into the telco DACS network(s), which subsequently
may be muxed into a fiber infrastructure along with traditional TDM voice traffic?
The DACS network can certainly be an overlay on a fiber infrastructure, and usually
is just that.

This kind of reminds me of a discussion I have had with several folks on a couple
of occasions on the issue of VPN's. To make a long story short, one of the important
roles of a VPN is to provide some sort of segmentation of traffic, and this can be
achieved at virtually any layer of the protocol stack -- even the physical layer.
An example of this sort of segmentation could be either a timing slot in a SONET
pipe or even a separate wavelength in a WDM system. And in most cases, at the lowest
levels, you have voice and data traffic riding the same light rails anyway, like
ships in the night. You get the idea.

Discussions on true optical networking aside, for the moment anyway.

It is no surprise that both voice & data services are affected by fiber cuts.

It begs the question then, if the Internet can work quite well today without
all the supposed telco 99.99999+, do we really need telco "reliability" -
SONET rings etc?

This kind of reminds me of Bullwinkle trying to pull a rabbit out of his hat.
"Nothing up my sleeve..... Presto!"

I don't think it is quite that straightforward. You first need to ask yourself,
"What problem am I trying to solve?"

As you know it is our contention that you don't need it. We believe that a
99.9999+ reliable Internet can be built without the underlying (supposedly
"relaible") telco infrastructure.

Define "reliable".

Well, my observation is that the principle difference in reliability between
the "traditional" voice network (PSTN) and "traditional" data services is
in how the underlying delivery services are fundamentally designed. This is
just a byproduct -- a fundamental difference in the underlying technologies --
circuit-switching vs. packet-switching. Each were designed with different
constraints in mind.

Let's not forget that IP is a connectionless, store-and-forward, datagram
delivery mechanism.

The remainder of this exercise is left to the reader.

- paul

Paul Ferguson said:

Funny you should mention this. Actually, aren't most data
circuits (excluding
copper pairs into the home) fed into the telco DACS network(s),
which subsequently
may be muxed into a fiber infrastructure along with traditional
TDM voice traffic?
The DACS network can certainly be an overlay on a fiber
infrastructure, and usually
is just that.

I don't recall the exact figure, but something like 50% of US circuits are
on SONET rings. So yes, the voice and data traffic are equally effected.
However, on older non-SONET networks the carriers have a "hot circuit
switch" technology that will switch voice circuits in the event of a fiber
cut. However, most data circuits do not have such protection.

Private data networks are severly affected by fiber cuts, but the Internet
is less affected because of its own intrinsic self healing protocols.

Define "reliable".

Good point. Many carriers and telco manufacturers are coming to the
Internet business saying that we have the knowledge and expertise to build
99.99999+ reliable networks - therefore eventually we are going to take over
the business. Reliable is easy to define in terms of voice calls. But it
has whole different set of meanings in the IP world, which most carriers and
telco manufacturers fail to understand. For example I can have a 99.9999+
SONET network, but if I can't access my DNS root server, then I clearly
don't have a reliable network.

Bill

Bingo.

- paul

Yes, the answer is to run your own root-server, a la GRS.

Actually, telco quality is not a positive thing at all in my mind:

- Circuits costing 40K/month take errors and are mysteriously "cleared
while testing" or better yet, there's "no trouble found" and the errors go
away all by themselves.
- MUX software upgrades that should have caused "several 50ms hits",
takes out 10Mbps of bandwidth for 12 hours.
- FOC dates that are more like guidlines.
- Having LECs and IXCs that finger point at each other's legs of a circuit
when there's a problem.
- Two words: unmanned terminal.

And it's not limited to the data services...
- Getting the infamous "all circuits are busy" or a fast busy when trying
to complete a call.
- The static filled line that used to be clear until the PacBell truck
came a prowlin.

Considering about 97% of our data network outages are caused by the telcos,
I shudder when I hear that Lucent wants to make my data network as reliable
as the voice network.

I can only hope that the current trend of realistic SLAs makes it to the
carriers, instead of the current "well if it's down for a couple of days
_maybe_ we'll refund some money to you if you beat us up really bad."

</rant> :slight_smile:

-Geoff

But that *is* the current SLA trend, and has been for years.

- paul

Data point. We recently had GTE erroneously delete a circuit so completely
that they couldnt find it in *any* of their records. Had to have a lineman
go out on site and physically trace the circuit back to the CO. Took them
12 hours to get the circuit back up.

"Telco quality" my ass.

-Dan

Heh. Try these:

http://www.forest.net/advanced/isp/gtesucks1.html

Regards,
Chris