Notes on the Internet for Bell Heads

Mailman List Mirror (Read Only)
1

Has anyone written the equivalent of the old Bell Systems Notes on the
Network for the Internet? A couple of books come close, Hueston's ISP
Survival Guide and Cisco's ISP Essentials. But there doesn't seem to
be anything that helps Bell heads understand what switching, routing
or signaling means on the Internet. There are a lot of words which are
spelled alike, but mean very different things in the Bell world and the
Internet world.

I've been thinking of it like driving in England or the USA. We drive
on different sides of the road. Its safe until you get someone who
doesn't know the rules of the road driving on the other side of the
Atlantic. So how do you explain the rules of the Internet road to someone
used to driving on the telephone system?

2

Hrmn, I can seem to download standards from
http://www.ietf.org/ just fine.

For some reason, I can't download anything from
http://telecom-info.telcordia.com/
all the documents seem to cost about $700 apiece.

Not to mention ANSI, ATIS, IEEE, ISO, ITU-T, TIA,
EIA, et al.

-dre

3

Actually, the reverse would be useful, as well. Voice Networking/SS7 stuff
for us IP weenies. (i.e. not voice over IP, just straight voice)

- Dan

4

Working for a Telco with an ISP division, I can tell you the best thing to
to do is wait for the Bell Heads to retire for the third time and keep
them away from your gear until then :slight_smile:

But in all seriousness, a book or set of documents would be very helpful
for those few Bell-shaped Heads that want to change their evil ways.

-Scott
(who is still trying to get back the IQ points lost in trying to
understand the SS7 network and being amazed that calls ever make it
through)

5

"Integrating Voice and Data Networks," Cisco Press, ISBN 1-57870-196-1

Part I, "Traditional Voice Networks" was a marvelous clue-by-four
for me as far as voice networks goes. I've read more books on the
topic than I can remember--which tells you how lousy the books
were. This one got me clued in fast.

I wouldn't claim to be a voice god now, but at least it was accurate
and in-depth enough to allow me to talk semi-intelligently with
those who live in that world ("E&M? Well, I'm not into that...but I
guess I can spank your ass if it'll get the circuit up any
quicker.")

Note to ciscopress.com: put the Table of Contents for your books
online!

Ch. 1 The State of Voice Communications
Ch. 2 Enterprise Telephony Signaling
  Signaling Functions, Analog Voice Trunks, Digital Trunk Types, R2
  Signaling
Ch. 3 SS7
Ch. 4 Call Routing and Dial Plans
Ch. 5 Defining and Measuring Voice Quality
Ch. 6 Voice Digitization and Coding

6

Working for a Telco with an ISP division, I can tell you the best thing to
to do is wait for the Bell Heads to retire for the third time and keep
them away from your gear until then :slight_smile:

Yes, several people mentioned that the two groups should just maintain
their seperate ways. There is this thing called convergence. If you
squint real hard MPLS can almost make an IP network look like a telephone
network. Add into the mix the government is desprately seeking ways to
make the Internet "secure." So many vendors are trying their darndest to
find a problem so they can sell a solution, even if that means creating
the problem in the first place.

I don't know which is scarier. Lucent/Bell Labs trying to design the next
generation Internet architecture, or Cisco trying to design the next
generation DCN/SS7 architecture.

(who is still trying to get back the IQ points lost in trying to
understand the SS7 network and being amazed that calls ever make it
through)

I'm certain the Bell heads are equally amazed that packets ever make it
through the Internet. The public telephone network is still the largest
network on the planet, and some amazing engineering went into creating it.
I'm not going to diss telco engineers. But a Babalfish to translate would
be useful.

How do you explain Internet security to a telco engineer. Or the concept
that the Internet doesn't have a LERG, but somehow ISPs figure out how to
get traffic from point A to point B. Or the biggie, that stuff is
expected to fail, so that's why you buy lots of simple, cheap ones instead
of one big, expensive, never-fail box.

7

I don't know which is scarier. Lucent/Bell Labs trying to design
the next generation Internet architecture, or Cisco trying to
design the next generation DCN/SS7 architecture.

the contest is keen. for a nice view of this insanity fueled by
greed, paranoia, greed, and oh greed, see the ieprep wg of the
ietf.

The public telephone network is still the largest network on the
planet

yup, and it transports less. is there a lesson about complexity in
that?

How do you explain Internet security to a telco engineer. Or the
concept that the Internet doesn't have a LERG, but somehow ISPs
figure out how to get traffic from point A to point B. Or the
biggie, that stuff is expected to fail, so that's why you buy
lots of simple, cheap ones instead of one big, expensive,
never-fail box.

dave meyer has been trying in draft-ymbk-arch-guidelines-03.txt,
and will give a bit of a talk on this at ie-piggy on sunday.

randy

8

Yes, several people mentioned that the two groups should just maintain
their seperate ways. There is this thing called convergence.

I know a small number of operators with really talented and dedicated
architecture people who have made converged networks work, and have
in consequence both reduced their costs and increased the number of
products they are able to offer.

I know way more operators with really talented and dedicated architecture
people who are preaching the gospel of convergence, and investing in
new equipment to support it, and are having their efforts sabotaged at
every turn by voice and data people who have closed ranks and are
defending their respective empires.

These operators wind up having to operate three networks (data, voice
and data+voice), with correspondingly increased operational costs.
The interop issues (both operational and architectural) between the
three networks increase complexity, reducing the chance that any
convergence products ever come to market, neatly and efficiently
defeating the entire point of the initial exercise.

How do you explain Internet security to a telco engineer.

You change the subject and make him feel good about his voice switches
until he wanders away and loses interest in bothering you.

Joe

9

http://www.bell-labs.com/news/features/jaffe.html

  Impact of Optical, MPLS
  According to Jaffe, advances in network architecture and technology now
  coming out of Lucent and other companies may have a profound impact on
  cyber-security in future networks deployed by both service providers and
  enterprises.

  "Three or four years ago, all-optical switching was considered science
  fiction, but Lucent is providing a path to that reality with the
  LambdaRouter," he said. "All-optical networks don't exist yet, but they
  are coming, and they will greatly reduce vulnerability.

  "It's very hard to intercept individual packets in an all-light network
  because they aren't queued in output buffers at intermediate nodes. And
  a lightwave network gives you a better idea of where the packets have come
  from, which is a problem with the Internet routing protocols today's
  data networks use."

I'm afraid this is one of those things I need help translating. I don't
understand how an all-optical network improves the security of the IP
layer. At best this is "improving" the security of the least vulnerable
part of the network. But I could be wrong, and I'm willing to be
educated.

10

Add into the mix the government is desprately seeking ways to
make the Internet "secure."

No, "control the internet"...security only applies to THEM....and their "big brother' intents...

So many vendors are trying their darndest to
find a problem so they can sell a solution, even if that means creating
the problem in the first place.

Hegelian principal in its essence....

11

It sounds an aweful lot like your initial analysis was dead
on. Unless, of course, we're talking about running glass
to the desktop again - the cost of which is going to be rather
prohibitive for most people.

And we're going to expend that cost for, as you pointed out,
one of the least vulnerable parts of the network.

That's rich.

12

They don't mention IP at all except by inference via MPLS.

http://www.mplsrc.com/faq1.shtml#MPLS%20History item C talks about migrating
layer 1 and 2 functions up to layer 3.

Maybe their assumption is that by supplanting IP at layer 3 in the
core, they will remove possible angles of attack? Of course, the flip
side of that coin is that they will create new ones.

I would imagine this looking somewhat like the IP stack only being used
at the desktop level. IP would be encapsulated within [insert all-optical
network protocol here] which would be used for the actual transport
and routing.

To take this thought experiment a little farther, in the world I describe
above, my ability to attack individual hosts is still roughly the same,
but my ability to attack the network itself has changed dramatically.

DOS attacks could be easily traced back to individual hosts and squelched,
maybe even automatically. With no global routing table to munge up, it
would be harder to black hole or flood.

Add in the fact that optical sniffing, while not impossible by any means
today, will increasingly become non-trivial as bandwidth increases. Which
is exactly one of the 'problems' they expect optical network to solve.

Regards,

Chris Kilbourn
Founder

13

Add in the fact that optical sniffing, while not impossible by any means
today, will increasingly become non-trivial as bandwidth increases. Which
is exactly one of the 'problems' they expect optical network to solve.

You mean just expensive, right? i.e. a couple transponders and an OC48 or
OC192 switch.

-Ralph

14

Cost is a factor, certainly, but the storage of the captured
data becomes the larger problem.

In the TB or PB range of optical data transmission, where and how do you
store the captured information? Unless you have TB's of solid state drives
to stream electrons into after an optoelectronic photon -> electron
conversion your only other option is to store the photons in loops of
fiber with an optical repeater.

Until we have quantum computers which might be able to parse the data in
real-time, we still need a buffer to store the data in before we can
look for the needle in the haystack.

Even with some nifty filtering on the sniffer, you're potentially
looking at obscenely large amounts of information to store.

I would expect that the distance of fiber you will need to store the
data in will be the gating factor, which means it tilts more towards a
physical issue than a cost issue.

If I need a few thousand kilometers of fiber as a storage loop, it's
kind of hard to move around efficiently. :slight_smile:

Regards,

Chris Kilbourn
Founder

15

Actually, research has been done that uses rare gasses to slow and even
stop the photons down in a tube. It would be possible to store the
states of photons in these tubs and then release them when you wanted
with out requiring miles of fiber. Also, photons work inpairs. It may
be possible to split the pairs on the fiber and observe the actions in
the fiber remotely by capturing one side of the pair and allowing the
others to continue. They interact in pairs even though physical
distance is between them.

16

Hello,

Since I am one of their customers I called and was told they will be
stopping their service in California. At this time they are not telling
me how much time I have before the serice is stopped.
Any pointers who supplies xDSL and will route my legacy and portable IP
space that was issued before ARIN was formed?

Thank you for choosing Sprint Business DSL service. It has been our
privilege to provide this service to you. Sprint announced today that
we will decommission our current high-speed data platform in cities
where Sprint has found a better way to deliver service via an
alternative platform or where high access and infrastructure costs make
it difficult to continue service.

Next week, we will provide you with pertinent information about
alternative high-speed service options. At that time, we will
communicate more information on how this announcement will affect your
current DSL service.

We thank you for your business and hope you will choose to continue as
our valued customer.

This is an auto-generated email, please do not reply to this address. If
you would like to speak with one of our customer care representatives,
please call 1-800-777-4686.

We will be in contact soon.

Thanks,

Michael

17

You might want to check out what the current prices are for a T-1. I used to work for a 9-12 school, and I upgraded their old frame 56k to sdsl. My friend who just took over the job is upgrading to a T-1 this summer. Not through the local telco it seems, but he says it will be the same price for the full T-1.

YMMV

Peter Hill

18

Actually, research has been done that uses rare gasses to slow and even
stop the photons down in a tube. It would be possible to store the
states of photons in these tubes and then release them when you wanted
with out requiring miles of fiber.

That process requires a reference beam, and currently can only be
tuned for specific wavelengths. Decoherence also occurs fairly rapidly
in the system.

In a DWDM network, you would need as many units as there are wavelengths
in order to do a full capture.

http://www.sciencenews.org/20010127/fob1.asp

Given how fast this field is moving though, it may become practical
sooner rather than later.

Also, photons work in pairs.
It may be possible to split the pairs on the fiber and observe the actions in
the fiber remotely by capturing one side of the pair and allowing the
others to continue. They interact in pairs even though physical
distance is between them.

I think you may be thinking about quantum-entangled pairs. That
phenomena is better suited to cryptography than general networking.

In an entangled system, both recipients would know pretty quickly that they
did not receive their photons as there would be an early 'measurement' on
one end, and a missing photon on the other.

19

Add in the fact that optical sniffing, while not impossible by any means
today, will increasingly become non-trivial as bandwidth increases. Which
is exactly one of the 'problems' they expect optical network to solve.

You mean just expensive, right? i.e. a couple transponders and an OC48 or
OC192 switch.

Depending on what you are trying to gather, it will also become more difficult at higher speeds to due the data volume. But you are right in that it's more about money than effort in the end.

- kurtis -

20

The discussion is certainly entertaining, but --

1) All-optical networking is a bunch of nonsense until optical processing
   ability includes complete set of logic and storage elements - i.e.
   achieving fully blown optical computing.

   Rationale for the statement: telecom is fundamentally a multiplexing
   game, and w/o stochastical multiplexing a network won't be able to
   achieve price/performance comparable to that of stochastically muxed
   network. Stochastical multiplexing requires logic and storage.

   The current opcial gates are all electrically-controlled, and either
   mechanical (and wear rather quickly, too, so you can't switch them
   per-packet or whatever), or iherently slow (liquid crystals), or
   potentially fast (poled LiNbO3 structures, for example) but requiring
   tens of kV per mm, making it slow to charge/discharge.

   Besides, your truly years ago invented a practical way to achieve
   nearly infinite switching capacity in electronics. Too bad, Pluris didn't
   survive the WorldCom scandal, as some investors suddenly got cold feet.

2) Wiretapping does not require storage of the entire traffic stream; and
   filtering for the target sessions can be done relatively easily at wire
   speed.

3) Nitpicking:

I think you may be thinking about quantum-entangled pairs. That
phenomena is better suited to cryptography than general networking.

In an entangled system, both recipients would know pretty quickly that they
did not receive their photons as there would be an early 'measurement' on
one end, and a missing photon on the other.

   You cannot detect "measurement" per se. What you get is skewed
   statistics; the entangled pairs obey Bell inequalities, which no
   classical system can. This gives an opportunity to detect insertion of
   anyting destroying entanglement of the pair - but only statistically.
   You need to send enough pairs to distinguish normal noise from intrusion
   reliably.

   Besides, quantum entanglement cannot be used to send any information at
   all. What it gives is the ability to get co-ordinated sets of
   measurements at the ends, but the actual results of those measurements
   are random. I.e. you can generate identical vectors of random bits at the
   ends, but cannot send any useful message across using only
   entanglement.

   Therefore quantum entanglement (aka Einstein-Podolsky-Rosen paradox)
   does not violate the central postulate of the special relativity theory (that
   no kind of entity can propagate faster than the speed of light in
   vacuum, in any non-accelerating reference frame).

--vadim