re: switching forums
Tito, the questions you've asked, despite their apparent simplicity,
requires that we adopt some common language for discussion. Your
apparent confusion is well taken, by the way, as the terms used to
describe the processes you are referring to have many different
connotations and meanings in both telecoms and elsewhere.
The terms "switching" and "routing" have been assimilated into the
language in much the same way as the apes did upstream, by mimicking
their forebears through inheritance and convenience. Everyone
understands that differences exist between a light switch, a voice switch,
and an IP packet router. Right? Let's see.
------
"What is considered switching/routing which the article refers to?.. Is
switching turning bits off and on. 1 for on, 0 for off."
Irrespective of the article for now, let's first talk about switching, and then
routing, in general terms. The term "switch" has many uses, as does route,
both in and out of telecomms and the physical arts, depending on the way
in which it is used.
"Or is switching: directing the channel to different directions?"
Not to seem facetious, but for both parts of this initial question, the
answer is 'yes.' It could be used to denote both.
-------------
There is first, due to its simplicity, the "light switch" model, which
carries over into many spaces in both electronics and photonics, on
substrate elements, cells, etc. This principle is responsible for simply
altering the states of electricity, and light, between the states of
on and off, or 1 and 0, by most accepted conventions..
Sometimes, in photonics, this also means altering the flow of photons from
high intensity to low intensity, as in the means that light streams alter in the
non-return-to-zero, or NRZ signaling format. But in this simplest of forms,
this is commonly referred to as the Light Switch model, regardless.
In semiconductors, this is achieved using transistors which form gates, and
most commonly associated with digital on-off or 1-0 state changes. In
optics, the same holds true, the difference being the nature of the flows
[electron versus photon] and the means of gating and biasing, or turning
on and off, those flows.
=====
In telecomms, when speaking of the higher layer attributes of switching,
instead of discrete electron or photon flows we speak about paths which
are turned on and off, or more to the point, established, by
switches. A switch, in this sense, might connect one link, say, (a --> b)
to another link, say, (c --> d) so that you end up with an end-to-end
connection, or a resultant path, between a --> d.
The switching entities that most often come to mind in telecoms are the
electronic switching systems, or ESSs which are found in your local telco
central office, generic cross-point switches, or matrix switches, like digital
cross connect systems (DCSs), such as those made by Tellabs, Alcatel,
LU, etc., T1 and SONET time division multiplexers, etc. This class of
switches are regarded as operating at Layer 2 in the OSI Reference
Model.
Routers, for lack of better terminology when they were first introduced,
are also switches. They switch a-b to c-d, too, and you once again wind
up with an a-d connection, but they do so under the direction of different
instruction sets: routing tables and policy criteria. Routers, in this case, are
considered Layer 3 devices. But in their root form, they are switches, too.
(I should note that router make ups have now gone back in some ways
from their original characteristics which were solely software-based in
nature, to include more of the simpler forms of silicon based switching,
too, by shifting some of the decision and forwarding processes to Layer
2. A case in point here is Cisco's Tag switching, now being adopted by
the IETF as mutli-protocol label switching, or MPLS. But I promised to
keep this simple...)
Adopt the following rule, and you can't go wrong: All devices that attach
one path to another path can be regarded as switches, in the telecom
sense.
[Sidebar: The gent who recently joined us here, Ed Edwards, made this
point to me back in 1989 when we were re-examining the merits of the
then Stratacom T1 packet engines, versus the then newbie Welfleet and
Cisco "routers," against the attributes of digital cross connects, made by T
and TLAB. He taught me to get out of the box on this topic when he
stated, in his own inimitable way, that: "They're all switches: A switch is a
switch." KISS, indeed. I'm still smarting from that one a bit, but he was
absolutely right. End Sidebar]
So, light switches (both electron and photon), electro- and opto-
telecomm switches, and routers, they are all switches.
The light switch variant takes place on the substrate, where streams are
born, and the telecomm path variant takes place at a higher layer (either 2
or 3), at the network element layer. KISS. Keeping Individual Switches
Separate.
=====
Take what I've laid out here as a step one, and re-apply it to your original
questions. Then return rephrasing your remaining issues, and we'll clear up
the rest.
Regards, Frank Coluccio |
