[Courses] [Networking] Lesson 3 and 7/8ths

Hamster hamster at hamsternet.org
Thu Jun 12 15:23:58 EST 2003


LESSON 3 and Seven Eighths

Well it's about time I gave the answer to the homework question for lesson
three. I'd also like to share a short anecdote and finally there's
something I'd like to bring up as a result from feedback from lesson 3.


1. Homework Answer

As a refresher, the diagram being referred to is:

<http://linux.invisiblepixels.org/writings/networkcourse/img/homework1.jpg>

The cable you need to link two hubs together is a crossover cable. The
reasoning behind it is quite simple. Recall the "Originate TX, Terminate Rx"
rule, and then follow the path of the packet leaving PC 1. The packet starts
off from the laptop and enters the hub on the Tx pair. The hub then performs
its voodoo and swaps the packet onto the Rx pair before it exits out the
cable joining the two hubs together. So far so good. But if we have a
straight through cable joining the two hubs, then the packet enters the
second hub still on the Rx pair. This is in clear violation of our rule.

However when we use a crossover cable, the problem is fixed. The packet
leaves hub 1 on the Rx port but enters hub 2 on the Tx port. All our rules
are obeyed and everyone is happy. The second hub performs more voodoo,
ensuring the packet leaves the second hub and enters the desktop computer
(PC 3) on the Rx port.

It might have sounded superfluous when I specified in the question "assuming
the hubs have only standard jacks". I did have a reason though. Some hubs
have a switch that allows you to change the wiring on one jack, thus
eliminating the need for a crossover cable. Some of the new and very fancy
ones even have a special auto-detect system that senses the type of cable
you've plugged in, and adjusts the internal wiring accordingly.

The picture below is of a hub with one of the switches I mentioned. The
"normal" setting provides a normal jack, the "cascade" performs a crossover
and allows you to join this hub to another one with a straight through
cable.

<http://linux.invisiblepixels.org/writings/networkcourse/img/hubcascade.jpg>


2. Anecdote!

Just a short story to hopefully give you all a laugh. I saw a network card
advertised on ebay that was exactly the same as the one pictured in lesson
three (ie with the RJ45, Coax and AUI connectors). The description of the
item proudly claimed the card had connections for "UTP cable, Coax cable
and a Joystick!".

The owner had mistaken the AUI port for a joystick port.

Well I though it was funny :-)


3. MAC address vs IP address.

I got an email saying "hey what's this about MAC addresses, I thought IP
addresses were how things knew where to go". Here follows another attempt to
explain the difference and why we need both.

It kind of reads as a bit of an abridged history lesson, cause it starts off
"in the beginning". So. In the beginning many different companies and
organisations began independently developing their own Local Area Networks.
Over time this led to a large of array of different networking technologies,
none of which could interact with each other. One of these networks was
called ethernet, and at its heart was the MAC address, a unique serial
number given to each individual network card.

At one point people began to think "gosh wouldn't it be nice if I could
communicate with computer X over in the next building, or with computer Y
two states over", but it wasn't possible with the existing network
infrastructure. So began the development of software to provide a delivery
and addressing scheme that could be used to enable all these different LANs
to communicate with each other.

Fast forward to the present day where out of the murky past the internet
emerges. The internet is nothing more than a huge collection of vastly
different networks all joined together by software common to everyone. That
software is TCP/IP, and is capable of running on an enormous variety of
hardware. TCP/IP doesn't care whether the device is a computer, a satellite,
a printer or a toaster. IP provides a uniform addressing scheme that enables
any device running IP to communicate with any other IP enabled device.

IP and MAC addresses interact in the following way. Each computer on your
ethernet LAN has an IP address and a MAC address. A programme will say "send
this info to the computer with IP address 1.2.3.4" and then passes that info
to your network card. Now your network card knows nothing about IP
addresses, but being ethernet, it does understand MAC addresses. So it finds
out the MAC address of the computer with IP address 1.2.3.4 and is able to
transmit the data.

In a peanut shell, IP addresses live at the software layer, MAC addresses at
the hardware layer. IP tells a packet where to go, MAC addresses actually
get it there.

We'll be having a closer look at this procedure in the next lesson.

Hamster


More information about the Courses mailing list