hpr-knowledge-base/hpr_transcripts/hpr0363.txt

Episode: 363
Title: HPR0363: Networking Basics Part 3
Source: https://hub.hackerpublicradio.org/ccdn.php?filename=/eps/hpr0363/hpr0363.mp3
Transcribed: 2025-10-07 18:57:42

---

.
Hi everyone. Thank you for listening to Hacker Public Radio. This is Clat 2 and this is
the third episode of my network series, which is basically just the bare bones basics of
how the concept of a network functions. In the first episode, if you'll recall, we talked
about the OSI model, which is just a framework of sort of how it all functions, how it works
together. Seven different layers in that. Second episode, we talked about the components
of a network, the hardware components like routers, switches, hubs, things like that. Recall
that hubs are basically amplifiers that take a signal and amplify them and repeat them
indiscriminately across the rest of the network. Routers, filter packets based on IP information,
things like that, switches, switch packets, and break up collision domains that do not
break up broadcast domains. And bridges, I don't really know anything about. Apparently
they're like switches though. Okay, so in this episode, we're talking about Ethernet.
And Ethernet, you know, if a network is the body, then Ethernet is the veins or something
like that. Ethernet is the, it's two things. It exists on the physical layer as the Ethernet
cards in your computer or the Ethernet ports in a device and the Ethernet cables that you
string from one device to the other. So that's the physical layer aspect of Ethernet. We'll
go into deeper, obviously, in a moment. And then there's the data link layer aspect of
Ethernet, which is where we start looking at what happens to this series of ones and zeros
as it goes from, from literally the data link layer to the physical layer, what gets added
and how does this packet develop? How does this data develop from a segment to a packet
to a frame to whatever? So we'll talk about both of those two aspects of Ethernet. First,
let's just get kind of down and dirty with the hardware stuff. So the specs of Ethernet
is that it is, well, first of all, as opposed to some other form of cabling, like coax,
apparently they used to use coax for networking and fiber optic. That's obviously a big, kind
of a big deal right now still. I mean, fiber optic is a lot more, you know, carries a lot
of data. So somewhere between those, you know, we've got Ethernet. And it is 802.3 technology.
So you've heard of like 802.11 stuff like that. This is 802.3. And according to various
think tank things, like electronic industry alliance and stuff like that, Ethernet is,
by definition, a registered jack with four five wiring sequence. So that is RJ45. So you
probably heard RJ45 being referred to as an Ethernet cable. This is why it's a registered
jack with four or five wiring sequence on unshielded twisted pair. That's UTP cabling. And
we'll talk a lot about the wiring within the Ethernet cables in a little while. The other
term you'll hear some people apparently throw around, although I haven't actually heard it in
person, is an eight pin modular connector. I read that that's what it was also being called.
I've never heard it called that, but there you go. Okay, so Ethernet cables, there is some
inherent attenuation, meaning that there will be a degradation of your signal strength.
The longer the length of the cable is. So you need to look, I don't know the maximum length,
you know, or whatever. And I don't know that there is, I'm sure there's a maximum, but I don't
know what the ideal length would be, but that's something that you have to decide based on your
network. So attenuation, like I say, it's the degradation of your signal. So that's measured
in decibels and it's rated in categories. So you might have seen when you're out shopping for
Ethernet cables. You might have seen cat three cables for a bargain price and then cat five
price a little bit higher. That's because cat five cables have more twists per foot of cable,
resulting in less crosstalk between the pairs of cables. So you've got, you know, you've got
a better signal. So if you can, get cat five cables and avoid the cat three cables. In my limited
experience, I couldn't even find cat three cables to save my life at the business I was working at
for a little while. This past summer, I was actually set on a task to get some cat three cables
to test a network on like really, really, you know, old and bad equipment. I just could not find
them. So I was feeling if you walk into a store and grab an Ethernet cable, I mean, as long as
an electronic store, you're probably grabbing a cat five cable. Like I say, cat three, you might be
able to find that bargain price is somewhere. So if you've got a situation where you're buying like
a lot of cables, like long, long lengths of Ethernet cable to set up a network for someone,
keep in mind that the length of the cable will matter in terms of how good your signal is and the
way to improve your signal as you get longer lengths of cable would be to do what? Well, yes,
to hook up a hub and that way you can have, or you know, you might call it a multi-port repeater,
and that way you've got your Ethernet cable going in. It amplifies the signal, gives it new life,
and sends it out across the cables to continue its networking destiny. Ethernet also uses something
called CSMA-slash-CD that is carry your synths multiple access with collision detection. Okay,
what does that mean? Well, each device on your network is going to, because this is Ethernet,
it can look on the wire to see if anyone else is transmitting a signal. If the path is clear,
your device that you're sitting at, you know, that you want to send a signal will send that signal.
Now, in the event that two devices transmit something and there is a collision, then Ethernet will detect
that there was a collision, and it will send a jam signal to everything on that network and say,
okay, we've got collision, data has been damaged, everyone put yourself on hold for a certain amount
of time, and then try to retransmit, and that's what happens. And this is all made possible,
because Ethernet has that functionality. I couldn't tell you which little wire it went across on
or anything like that. I haven't done that much study on Ethernet, but that is something that is
a feature of Ethernet. And that's a good thing, right? Because you've got collision detection,
and you've got a way to detect whether there's a signal on your wire already, and you know,
you can wait to send your signal. I mean, you, the user, wouldn't do that, but the software,
the networking stacks underneath your applications and things will do that. And that is a good thing,
but keep in mind, you know, if we're talking about a home network with, you know, six or ten computers,
not that big of a deal, but if we're talking about something that is, you know, a lot more complex,
like your employer needs you to set up a big network for all of the employees and the company
and stuff, then you might want to take a look at that stuff and really think about breaking up
your collision demands, because otherwise you're going to start having people trying to send out,
you know, whatever signal a request for a web page or an email or whatever, and there are computers
just going to put itself on hold because the line is busy right now, and then it's going to keep
holding because, you know, you've got 200 other people trying to send signals out over this
network, and that is why we like to break up our collision demands to avoid issues like that.
There are two types of Ethernet cables broadly speaking, or actually there's, it's not just Ethernet
cables, it's like Ethernet interfacing, it's half duplex and full duplex. There are important
differences between the two, and you're going to find equipment that is capable of both
and equipment that is capable of only half duplex. Again, I had an old job over the summer that I
was working, they assigned me a job. They sent me the task of testing out a network on half duplex
with old cables and all this other good stuff, and I mean literally I had a hard time finding a switch
that would not do full duplex, so it was really tough to find something that, you know,
that wouldn't do the faster of the two things, so chances are if you're working at a place with
fairly modern equipment, then you're probably not going to have to worry about this, but for some
reason you're working at a place with old equipment and you're experiencing a slow network,
and you're trying to figure out what's going on, where the bottleneck is, whatever. That's one
thing the check can be aware of, that certain, you know, every element on your network needs to be
capable of full duplex from maximum speed, because half duplex is sort of a one-way street for
networking. There's one pair of wires, and all the data is being sent via this pair.
Full duplex is a two-way street. It's got one pair of wires to send data and one pair of wires
to receive data, so in theory that would boost your network performance by like 100%.
So the full duplex will speed up your network. It's great for all your connections from, you know,
a switch to a host, a switch to a switch, a host to a host as long as you're using crossover,
and we'll talk about crossover in a moment. You do need a dedicated switch port for each node
that you want to be true full duplex, and you need to obviously ensure, like I say, that your network
cards and your switches, and the cables are all capable of full duplex, that you're not trying to
use a half duplex cable on a full duplex switch, although I've done that and it seemed to perform
really well, but I wouldn't swear that that made it true full duplex. Ethernet comes in a variety
of different flavors. Three of them I've actually had experience with, and so I'm not going to try to
go into any of the other ones, but it's just mostly important to understand, or to realize,
I guess, that all Ethernet is not necessarily the same. I mean, so there's 802.3, which I guess
originally it was like a three-megabit per second connection, but now everyone, when you talk
about Ethernet and you're just talking about, you know, the slow Ethernet, you would be talking
about 10 megabits per second. Above that, there's the 802.3 U, which is considered fast Ethernet,
which is 100 megabits per second, and then the step up from there is 802.3 AB, which is the gigabit
Ethernet on cat5 cables, and so those are the different speeds of Ethernet. You can look up 802.3,
or just like fast Ethernet, or Ethernet, like on Wikipedia or something, to find out all the
different variations of that. It's really something that you'll probably want to look up more on a
basis of should you ever encounter it, because typically, or at least in my very limited experience,
the thing that I had to know about was whether to have my switch set at Ethernet, 100 megabits
per second, or gigabit Ethernet. There was actually a physical button on the switches that I was
dealing with at this one job, and it was just a matter of having all of my switches set to
the correct thing. If you don't keep it in mind, obviously, there's a danger of you possibly
not having things set correctly, and not having an optimized network. Those are the different
types of different speeds that Ethernet that you have available to you. There are different kinds
of Ethernet cables that you're going to encounter as well, so you're going to want to make sure
that you've got the right one. There is a straight-through cable, there's a crossover cable,
and there's a rolled cable. The straight-through is going to be good for your host to switch,
or your host to hub connection, or your router to switch, or your router to hub connection,
and that uses four wires, full duplex, and it is wired in such, like, if you took the cable apart,
if you took the connectors off the end of the cable, you would see that the wires, you know,
the red, blue, yellow, whatever, those little wires inside the cable pin one connects to pin one,
pin two connects to pin two, pin three connects to pin three, pin six connects to pin six.
So if you took the ends off of each cable and examined it, you would see that the red cable
goes to the same place on the other end as it started out on the first end. So that's just
straight through. Now, crossover is a little bit different. It's going to be good for a switch to
switch connection, or a hub to hub connection, or a host to host connection, or a hub to switch
connection, or a router to host connection, and that also uses four wires, but it is wired in such a
way that pin one on, let's say, the left connection goes to the pin three on the right hand
connection. Pin two, if you follow that through the cable, you'll see that that connects with
pin six over on the other end. Pin three would go back up to one, and then pin six would go back
up to two. So obviously, there's literally a crossing of the wires there so that they end up
on opposite pins on each side of your cable. But anyway, the point is that if you're in a pinch,
you need something other than what you've got. Keep in mind that if you just look this kind of stuff
up, you can actually rewire it pretty easily yourself as long as you've got the right tools for it.
Okay, the last kind of cable is a rolled cable, and this is basically, it's really just used,
it's for a very specific purpose. It's when you want to connect a host to a router,
but via a serial port. So this is when you've got a fancy router that has a serial port,
you can use this rolled cable to connect from your host to that router via this port,
and that enables you to top directly to the firmware is the typical use for that. I've never
used one, so I can't really say much more about it than that, except that it exists, and that's
what it does. Looking at the cables themselves a little bit further, we should start thinking about
how data gets across these mysterious Ethernets. How does the data get from, for instance,
the application layer, and then all the way down to the physical layer? That's kind of,
it's kind of a good thing to keep in mind because it's actually, I mean, it's something that you're
going to be thinking about a lot in a network, because what happens is if you've ever seen a
space shuttle lift off on TV or whatever, you'll see that the space shuttle leave the ground,
and then once it reaches a certain altitude and it's burned off a certain amount of fuel,
that fuel tank will be ejected from the shuttle itself, so it's shedding off these layers that
it doesn't need at, you know, at different stages during its journey. It's kind of similar on a
network in a way. So let's say that we start out, obviously, at the application layer, where the
user is making some kind of request or generating some kind of data. The presentation layer deals with
that data, puts it into whatever format it needs to be put into, and the session layer below that
establishes a session as it were to keep all the data organized and make sure that the data comes
through where it needs to come through and ends up where it needs to end up, then the transport
layer comes into play, and it will add a header, either a TCP or a UDP header, to the data that's
being fed to it from these upper layers, and it also creates a virtual circuit for this session,
giving it an arbitrary port number, which would start at 1024 and go up from there. So let's say we're
trying to FTP somewhere. That is by default assigned to port 21, right? So if we're sending data to
a port called 21, and the computer sees you wouldn't want to send it from a port also called 21,
now would you? Or you wouldn't want to send it from a port that is assigned to some other service
already, like SSH. You want to give it a random, the transport layer is giving it a random,
or I don't know how random it is, but an arbitrary port number that is not reserved for something else,
and that is not going to be the same as what you're trying to communicate to. The network layer
then is going to add an IP header to the segment of data, and at this point this segment of data
becomes a packet. It's also going to add a protocol field defining whether it is TCP or UDP.
It's going to find the destination hardware address, and the way it's going to do this is with
something called address resolution protocol, which is otherwise known as ARP, and we'll talk about
that more on some other episode, but basically it compares the IP address and the subnet mask of
the source to the IP address and subnet mask of the destination. And if they're on the same network,
then ARP is going to ask for the MAC address, and it'll send the packet if it is not on the same
network. Then IP is going to look for the IP address of the default gateway or the router instead,
so that packet can be forwarded to the appropriate network. In other words, if it's not on your local
area network, it must be on some other network somewhere else on the internet, and it will send it
out to that other place, or not necessarily the internet, but it could be another network that
you're connected to. The data link layer then comes in and encapsulates each packet into what's
called a frame, and it places another header. This time of the source MAC address and the destination
MAC address. MAC address being, of course, media access control. That's the number that's
burned into the network card, the CRC, which is the cyclic redundancy check, which is going to
help the data link layer detect errors, although not correct them. That's added to the header as well,
and the answer to that CRC is placed in the FCS, which is the frame check sequence field
in the at the tail of the frame. So we've got a lot of data being added there, and then finally,
all that stuff is sent over to the physical layer as a bunch of ones and zeros, a pure binary
signal, and it's committed down to the network medium. That is the wires, the ethernet cables
themselves, and every device on the network synchronizes with the frames clock, and they extract
the ones and zeros from the signal. They build a frame, and they take a look at the CRC,
make sure that the frame is okay, and if it is okay, then the device checks to see if that frame
is for them. If the frame is for them, then the process continues, and it does what it needs to do,
kind of the reverse process of all that, it goes through all the headers and figures out what
protocol it is, and it makes sure that there is no, no or no dropped packets, etc., etc., and then
it sends it on up through the session layer to the application layer. If it's not for them,
then they ignore it, and it's not for them, so they don't worry about it, but that's how this
stuff is sent over the network. It's just, it's given a whole bunch of different headers,
lots of different information is being put onto the actual data packet that you are sending out,
so that all the different components of the of the network stack can, can take a look at it and
know what to do with it. So let's talk about ethernet as a data link layer entity. Ethernet addressing
uses the media access control number, and that is, like I say, it's burned into the ethernet
interface of a device. It is a 48-bit or six-byte number in hexadecimal format, and it's very
specific as to how this is composed. The first bit is the individual group bit. If it is zero,
then you know it's a MAC address. If it's a one, then it could be something else like a broad
cast or a multi-cast address. The next bit is the global or the local bit, sometimes also called
the universal bit, and if that's a zero, then this this is a globally administered address by the
IEEE organization. If it's a one, then it's a locally administered address. So if you are,
if you've got your own protocol or if you're running tests or whatever, I mean I don't think this
would be on the level of certainly what I would be doing, but if you were really screwing around
with the networking and stuff like that, if this was set to a one, then you would set this bit to one
because it is a locally administered address. The next 24 bits are a manufacturer's assigned code,
or again it could be a locally administered one, but assuming that this is more common, it's
going to be a manufacturer's assigned code, and this is usually just, it'll start with zero,
you know, 24 zeros for the first card that they churn out of the factory on up to the end of the
run, you know, whatever that would be mathematically to fit into 24 places. So that would be what you
would see there, and a lot of times that's the you'll see those same digits as part of the serial
number of the actual interface card as well. The Ethernet frames themselves are used to encapsulate
packets coming in from the network layer. So right above the data link layer obviously is the
network layer, right? They're done in a Mac frame format, and they provide error detection
from cyclic redundancy checks, the CRCs. So this is important data because it's double checking
essentially the integrity of the data. And so again, they're very specific in what they contain.
They've got a preamble, which is alternating ones and zeros, which create basically a five-meter
hertz clock at the start of every packet. This lets all the receiving devices on your network synchronize
to the incoming data and make sure that they're getting all the data. And the way it does this is
with what's next in the in the data frame is it's the start frame delimiter FDF, and that's
it's one octet. So it's one zero, one zero, one zero, one one. So everything knows, you know, we
can come in at any point during that one zero, one zero, one zero, but then when they hit one one,
they know they're in sync. The destination address comes next, and that could be an individual
address, or it could be a broadcast address, or a multi-cast address, you know, to all the
devices on your network, or just a subset of all the devices, whatever. That's the destination
address. So the next thing would be a source address, and obviously that's just going to be from
one thing that's not going to be a broadcast, or a multi-cast address. That's going to be an actual
48-bit MAC address of the computer or the device, the node that sent the data. After that,
is the length or the type, and that identifies the network layer protocol that's being used,
and then the actual data itself. So this is the data that you are sending, and that can be anywhere
from 64 to 1500 bytes. And then there's the frame sequence check. Like we said, that's the
that's the thing that's going to have the answer to the CRC, so that once the frame is all received
and looked at, the cyclical redundancy check can be verified by the frame check sequence.
That is everything that that composes the data that is being sent from the data link layer over
the Ethernet cables. And that about sums up Ethernet for you. So you've got it on the physical layer,
you've got the wires in the cables, different speeds, different, you know, full duplex,
the half duplex, you've got straight through, you've got crossover, and you understand the process
of data getting from the application layer down to the physical layer, and you also know what
exactly is being sent from the data link layer to the physical layer. So in the next episode,
we'll continue our exploration of networking, and thanks for listening.