This question is a little bit in broken English, but I’m going to read to you
verbatim:
“How signals travel by cable? One by one or side by side? From LAN, many
computers send and receive packets on a single telephone line. Whether they
travel one after another or in lanes?”
In this excerpt from
Answercast #57, I look at the way that data travels along network wires.
Become a Patron of Ask Leo! and go ad-free!
How does data travel?
So, first of all – it’s not a telephone line. It’s an Ethernet connection.
When computers are on a LAN, the plugs may look kind of like a phone plug,
but they’re actually a little bit bigger and it’s a completely different wiring
scheme. A telephone works on exactly two wires, whereas Ethernet actually
requires all eight connectors in that wire to be operational.
Now, with that little trivia out of the way, it’s interesting to note that
the answer is surprisingly “one by one.” And in fact, there are no lanes.
Well… there’s two lanes, I should say. On an Ethernet connection,
there are in fact two lanes: a lane going out (in other words, a connection for
transmitted data) and then one coming back (a connection for received
data.)
Data collides
When you get those all connected together (in say, a hub or a switch or a
router), everybody’s sending and receiving data at the same time, and in fact,
they collide.
It is very common. A lot of the Ethernet communications protocol is all
about detecting and handling collisions. The reason you and I don’t notice this
is that it happens so gosh darn quickly.
Packets themselves (once they get started, once you start transmitting a
packet) are a very short burst of information. The system finds out if there’s
a collision.
In other words… Two people try to talk at the same time (or to use your
terminology, two computers tried to occupy the same lane at the same time) and
unlike an automobile collision, the computers can simply try again. And in
fact, they will even try again at different times. They’ll delay.
If you think about it: if you had two identical computers running identical
software following identical rules, if they collided, they would follow
identical paths to try again, which would likely cause them to collide again.
So, built into this protocol are things like a certain amount of randomization
that says, “OK, I’m going to wait this much and that guy’s going to wait that
much.” And as a result, the packets will then get sent at different times and not
collide.
There’s one pipe
But ultimately, the answer to your question is there’s really one pipe. And
that one pipe carries all of the information from all of the computers on a
network segment.
Network segments can be isolated from one another. In other words if you
cross a router, one side of a router may have a completely different set of
traffic than the other. But when you take a look at a simple network connected
by a hub (or several computers that are connected somehow directly together), it
really does boil down to a single lane of traffic and everybody just sort of
collides like crazy, recovers, and tries again.
Next from Answercast 57 – If an open Wi-Fi hotspot displays a page asking me to login, does that make it secure?
To amplify on Leo’s answer, Ethernet is a CSMA system, ie Collision-Sensing, Multiple Access – and note the sequence of that title.
That is, the CS – Collision Sensing, is the more important aspect regarding getting access.
In general, it resembles accessing a Motorway (UK), Autoroute (France) or Inter-State Highway (USA).
Each driver looks out for and if necessary avoids other vehicles.
The other main system is Token Ring, which resembles railway/chemin de fer/railroad practice.
The driver/ingeneur/engineer is given a unique token or staff, then being the only one authorised to access the signal block or designated length of track.
There can be slight variations of this. The driver is shown the token at the entry signal box/tower, thus knows that the track section should be clear of opposing traffic.
This can be repeated for a number of trains, until the last driver is given the token, to carry through to the far end of section, where it is handed over to the exit signaller, who can then allow a train or trains to use that same section in the opposite direction etc.
The token itself is very clearly marked identifying it with the track section or block involved.
There are modernised version of this, involving radio tokens etc; but the principle remains the same.
When DSL has been explained to the layman it has been related to a ‘sideband’. You have to filter it to keep it out of your voice line. When you use a modem it takes the entire voice line.
While you do use the extra wires they still have to be converted to a hybrid form to travel on that voice network, once again to be separated for the ISP.
At least, that’s how I understand it and my gateway, and I’ve been using computers since 1979 and online for more than 15 years. When the gateway receives the line signal it splits it off for a clean voice line but the extensions must be filtered.
O.K., a slightly different topic…
My old printer has a Centronics connection – a Parallel Port. My old computer also had a serial port. Both have now been replaced by USB – Universal SERIAL Bus.
*I* thought parallel connections were faster, with 8 wires, each carrying one byte, i.e. all 8 carried one byte, whereas with a serial connection, the signal went down one wire a bit at a time, so took eight times as long.
Now I’m lost…..
As my grandmother used to say, they don’t make ink as black as they used to – and the bulbs aren’t as bright, either…
05-Oct-2012