All language subtitles for 09 - Switching - Welcome to the World of Switching!-eng

>> In the realm of networks you'll find there's
all kinds of devices out there, firewalls,

intrusion prevention system, NAS --

I mean there's all kinds of stuff
that can plug into a network.

However, there's two core devices that really
make it work -- switches and routers --

both of which Cisco has staked their name on.

They will tell you, "We make the best
switch and router than money can buy,"

and I will tell you I absolutely agree.

It takes a lot of money, and it is the best.

So with that being said, switches is where most
people get their start into the Cisco world

because they take a lot more normal day
to day stuff, normal day to day changes,

as compared to routers that a lot of time you
just set up and forget about unless you are

in a very dynamic and expanding organization.

So switches are a lot of your day-to-day.

So we'll start there, here.

We'll start there, here.

Did you get that?

So we'll look at the different devices, kind of
the evolutions that switches have gone through.

A big one for your understanding

and certification purpose is
collision and broadcast domains.

And then I want to walk you

through an end-to-end scenario,
a day in the life of a switch.

So the realm is 1980s' timeframe, Bengals
are singing Walk Like an Egyptian,

big hair is the norm, and network
hubs are hitting the market.

10 megabits per second speed is
screaming, this is life-changing events,

where we now can have computers sharing data

without people walking floppy disks
back and forth down the aisles.

I mean, that's the kind of
timeframe that we're in.

Ethernet is developing the standard of carrier
sense -- I should say it is developed --

carrier sense multiple access collision
detection, meaning we have this competition

between token ring, which is still a valid
competition back in the '80s, token ring

and Ethernet, where Ethernet
uses this kind of environment,

multiple people are accessing
the network at a time.

They're all listening.

They're all trying to sense if there's
anything being sent on the network.

If they don't hear anything, because
only, and here's a key point,

only one of them can send data
or receive data at a time.

Only one. And so they're all listening, they're
like, "Okay, is anybody sending any data?

No, I don't hear anything, so I will go ahead
and send it," and that works most of the time.

However, there is a chance that two
people are listening at the same time.

What if two computers are listening at the same
time, they both hear a clear, they both send.

That's where we have a collision.

And that is collision detection, that's the
ability for the devices to be like, "Oh,

we both sent at the same time,
our data is scrambled, my bad."

They run a back-off algorithm that they both
back off and then wait a certain amount of time

to send again and then they both send and
hope they don't collide a second time.

Because collisions are really taking
down their network performance.

And that was one of the things with token ring.

Token ring said we're going to send
this little ring around the network.

I shouldn't say send a ring.

Send this token around the network
that all the computers are plugged into

and only whoever has the
token can send at a time.

So this token is going bzz, bzz, bzz,
you know, flying at the speed of light

around this ring network, the devices
are grabbing, adding their data.

It never had a collision.

That was carrier sense multiple
access collision avoidance,

which you might say, "Well, that's gone," right?

So token ring's gone.

Well, yes, token ring is gone for the most part.

However, collision avoidance
has reared its ugly head again.

I shouldn't say ugly head.

It's there in Wi-Fi.

Wireless networks don't have the ability
to detect collisions, so they've gone back

to a type of collision avoidance
system, but that's a total other topic.

So hubs are sitting here in
Ethernet running in such a sense

that they have only one person
able to send at a time.

Now, let me put a definition to that.

That is short for one-collision domain.

Key topic to know: a hub, no matter how many
ports it has -- it could have, what is that,

eight ports like I'm staring at,
or 20 or 950,000 ports on a hub...

it would not have that, but it
would all be one-collision domain

which in rough English would just be how
many people can send or receive at a time.

Now, I also want to emphasize the "or" there.

"Or receive" means you are
in a half-duplex world.

Send or receive, so I'm pretty
much saying, "Well,

I can send something or I can receive something.

I cannot do both."

It's like a walkie-talkie.

Now, that was the world of the hub: one
person sending or receiving at a time.

So the larger and larger your network group,
the more and more collisions you would have

because there's more of a chance that people are
sending and receiving, or sending or receiving,

at the same time and you're getting collisions,
your network performance is going down.

Likewise, you run into challenges with security,
meaning a hub, if you bring up the OSI model.

You know, you've got your seven layers,
dut, dut, dut, dut, dut, dut, dut.

Down here is the physical layer,
which is where the hub resides.

Physical layer devices have
no intelligence at all.

They receive electric signals
and they send electric signals,

and that's exactly what this does.

When this guy sends a packet -- let's
say it's destined for this guy --

what the hub does is receive it and
say, "Well, I just got some data.

I'm going to send it out
all of the network ports."

This guy will get it, but so will
this guy, and so will this guy,

and whatever other guys are
attached to that network.

Now, he's got the sad face because
his monitor is not a perfect square,

but also because he is a hacker.

He is using a program like Wire Shark.

It used to be called Ethereal,
which you can freely download.

It's like those programs we just
talked about in the Council Connection.

You just type in Wire Shark in Google and
download it, and what it will do is capture all

of the data that it is receiving
on that network port

and store it so you can actually look at it.

If you were good at Wire Shark, which it
takes a little training, but not much.

Wait a sec, I've shown you Wire Shark.

What am I talking about?

You know what Wire Shark is all about.

You can reassemble Excel spreadsheets
that are being transferred.

You can capture voice conversations
that are happening, record phone calls

that are being used by voice over --
you can see emails that were being sent.

I mean, everything is able to be seen
because a hub sends everything everywhere.

So today -- that's enough about
hubs because they are network death.

You do not or should not use these
in production because, number one,

the collisions alone will slow your network to
a crawl; and second off, the security, I mean,

it's just, this is yesteryear technology.

Also a hub maxed out at 100 megabits per second.

Once it reached that speed, that's where
everybody's like, "Okay, we've gone switching."

You won't find a gigabit hub.

So we move into the 1990s.

The Bengals are now not singing
Walk Like an Egyptian.

Actually, it's really funny.

The Bengals performed at a Cisco live event.

You know, Cisco's big, once-a-year
conference here in the United States.

And they sang Walk Like an
Egyptian and they forgot the words.

They got to like halfway through
and the girl was singing --

you know, it's a really fast song.

It's hard to keep up with and she's like,
"I just totally," they forgot the words

to their own song and they made a joke about it.

It was funny.

Nonetheless, the Bengals are
forgetting their own lyrics,

we're coming out with network bridges in 1990s.

The network bridges was this transitionary
device between the hubs and switches

which broke the network into
multiple-collision domains.

Now, looking at it -- I couldn't
find a picture of a network bridge.

Looking at it you can't really tell much of a
difference other than bridges had limited ports.

You wouldn't find a 32- or 48- or 64-port
bridge that you would plug your devices into.

These were really expensive devices
that you would have your hubs,

with all your devices attached, and as it
started maxing out, like you're getting a lot

of collision, you would introduce a
bridge which would have maybe two,

maybe three, four different ports on it.

And that bridge...

would separate the multiple hub-based domains.

Remember, over here, only one
person can send at a time.

So a bridge would, number one, introduce
more than one-collision domains.

Every port on a bridge is a collision domain.

So over here we can have one
person sending or receiving.

Over here we can have one
person sending or receiving.

So now we can have multiple people sending
or receiving at a time on the network,

and the bridge is now introducing intelligence.

OSI model, we're now moving up to layer
two, we're at the data link layer,

which is where this device resides,
and it learned the MAC addresses.

So the hub, we've got, we'll say, 10
computers over here and 10 computers over here.

This bridge, as the computers are sending and
receiving, it's learning which MAC addresses

or which data link layer addresses
are on each side of the network.

So now you get some limited filtering.

When somebody sends something that belongs
on, let's say, this side of the network.

Let's say this guy sends.

It hits the hub.

The hub sends it everywhere, including
the bridge, and the bridge goes, "Oh,

well it looks like that was sent to this
MAC address," we'll call it MAC One.

"That was sent to this MAC address over here.

So I'm not going to forward that
on to this side of the network.

They don't need it."

So that's where our filtering came into play.

So now, if you sent to the
other side of the network --

let's say we sent from the left side of
the network here over to the right --

as soon as it hits this hub, it
explodes and goes everywhere,

so you're not really getting much filtering

since we still have these layer
one devices, but it was good.

I mean, this was a good step.

Now, here is the big Achilles
heel of the bridges.

They were software-based.

So these guys would slow down your network.

When I first started teaching,
I actually started with Novell

and then moved into Microsoft technology.

Talk about Windows server
MCSE certification back then.

And one of the things, I still remember this for
some reason, I was talking about how Windows,

Windows 2000 I think it was, could do RAID.

Meaning it could do RAID
level one or RAID level five,

which is mirroring two hard drives together.

And I had a guy in class go "Ah, ha, ha, yeah...

like I would do that with Windows."

And back then I didn't know better.

I was like, "Well, why wouldn't
you do that with Windows?"

He says, "No, no, no.

We buy separate hardware to do that.

We buy RAID controller cards,"
which are now common everywhere.

RAID controller cards that offload that,

so Windows doesn't have to
worry about mirroring those.

Well, in the same sense, Windows would slow down
a lot if you actually tried to do RAID with it

because it's having to keep
up with hard drive functions.

Same thing here.

The bridge is learning all these
MAC addresses in the software.

It's processing them in the software, so
as stuff goes from one side of the network

to the other, it slows down because of the
processing that's taking place on that device.

And now we make the turn
into the new millennium.

Around year 2000, you start seeing network
switches becoming a commonplace thing.

And it's funny.

I'm thinking "Okay, we had
the '80s with the Bengals.

What do you say happened around
the turn of the generation?"

You know, it's like the Y2K scare.

Well, that's an event.

There's nothing that really characterizes
things that have happened, you know,

maybe the dissolution of normal relationships
to where now everybody communicates via texting

and Facebook and no longer
face-to-face because that's weird

to actually see somebody and talk to them.

I'm saying that's weird to
have natural relationships.

Here I am talking to myself
staring at a screen, yeah.

I'm a lot better, right?

So anyway, here we are in this switching world.

Every port now has its own
collision, and you know what?

Let me actually move this to the top.

Because everything that I said about bridges
is the same when it comes to switches; however,

we now have a very high port density.

Like a lot of devices can plug into these.

We're no longer limited to
two or three or four ports.

But the biggest thing that
changed was the creation of ASICs.

What are those?

Application-specific integrated circuitry.

Such a huge concept for our
network devices, because this moved

that software-based processing
of the bridge into the hardware.

And you're going to see this topic
come up again and again and again.

All ASICs are is somebody engineered a chip.

Here's my little chip, a little chip with wires
and connections and all that kind of stuff

to where in the hardware the intelligence
of the device, it's able to handle it

without relying on any kind of software.

And any time you introduce ASICs into
the picture, you introduce a lot of speed

because you have it all being
processed in hardware.

No longer does the software
have to get involved.

ASICS has changed the world of
switching or bridging, I should say,

to where now switching is commonplace.

ASICs changed the world of VPNs.

We would no longer be able to
scale virtual private networks,

which is encrypted tunnels running across
the internet, if we didn't have these ASICs,

these chips that handle a lot of the encryption,

because the encryption alone would
bury the software of the device.

So ASICS are a world-changing event.

They cause a lot of these
devices to just move a lot faster.

Now we have switches where it's able to move
as fast as the hubs were, which is wire speed.

As electric signals are coming in,
it's processing and sending them out.

Every port on a switch is
its own collision domain.

So when you come to a switch, you
look at how many ports there are.

Let's say it's a 24-port switch
or this one looks like a 20...

they got these two weird ports,
so maybe a 26-port switch,

so you've got 26-collision domains.

And again, going back to our simple definition,
26 people, 26 devices plugged in there can send

and -- ooh, there's a big change
-- and receive at the same time.

We've gone full duplex to where now instead
of a walkie-talkie, you're like a telephone

to where both people can talk on top
of each other and still understand,

at least in the computer world,
still understand each other.

So when you say this is a 100-megabit per
second switch, or 100-megabit per second port,

really you're getting 200 megabits per
second if you're running it at full duplex

because you get 100 send and 100 receive
that you can do at the same time.

Now, all network speeds are rated at
half duplex, so when you see a gigabit-

or a 100-megabit per second port, it's
being rated at half duplex, so I mean,

you truly can never go above that speed.

It's just now you can do send
and receive at the same time.

Totally kind of life-changing event there.

You also support varying port speeds.

A hub had to have all the same speed.

If it was a 10-megabit per second port, every
port on that hub, a 10-megabit per second hub,

every port on the hub would
be 10-megabits per second.

But with switches you could have, for
instance, these guys, and this is common,

over on the left being 100-megabit per
second connection and these two guys

over on the right might be 1,000 megabits per
second or a gigabit, or as technology continues

to evolve, we're going to
see these being common,

1,000-megabit per second ports is
normal customer plug-ins and normal use

and then maybe these uplinks are
10,000 megabits per second, or 10 gig,

or now 40,000 megabits per second or 40
gigabits per second uplinks that you can have.

So that way you can have switches that are
daisy-chained together to where, you know,

you've got all these guys that
are all 100 and this is 1,000,

so that way this doesn't become a bottleneck
and all these guys are 100 and this is 1,000,

so that way you can link these together
without bottlenecking them on these speeds.

These switches are managed and intelligent
to where -- and I'll flip that term.

They're intelligent in the sense
that they have the same capabilities

of the bridge to learn MAC addresses.

So as these switches power on, they will learn
all the different devices that are on there.

So now when this guy sends a packet into the
switch, it will be able to send it out to just

that guy because he knows
where the MAC address is at.

He's located all of it.

He will even learn, for instance, if
you start daisy-chaining them like this,

you've got 20 MAC addresses down
here, we'll say, that it's learned.

This switch will learn that all 20 of those
MAC addresses are accessible on this one port.

So it's not uncommon to see one port with
20 MAC addresses, which tells you, "Okay,

that port must be daisy-chained
to another device

that has all kinds of devices plugged into it."

So that's the intelligent side.

The managed side is going to be
what we get into with the Cisco iOS.

We can do things with our switches.

We can change things.

We can modify settings.

We can add features.

Now, one of the nice things about Cisco
switches, out of the box, they work.

That's great.

You plug them in, you plug in your
computers, and it's working like a switch,

but you're only getting limited
feature set by doing that.

You're just getting the base functionality.

You might as well go buy some other brand
or what you'll hear is an unmanaged switch,

with doesn't really have the iOS.

It doesn't have features.

You just plug it in and plug stuff
together and it just kind of works.

It learns MAC addresses and does its thing.

Cisco adds features that you can configure.

That's going to be what we talk
about in the switching world today.

Now, I know you're looking at, I was just
thinking of what else about these things?

They smell good and they're smooth
-- now, I'm only half joking.

When you open, if you have the
privilege of opening a Cisco switch,

it has that new cellophane smell.

It only lasts for about 10 seconds
after you pull it out of the box.

It's so good.

After you've opened a number of these
devices, you know, initially it's like "Ooh,

that's repulsive," but after a little while
you're like, "Man, I just, I love that."

It's like Pavlov's dogs, you
know, you kind of jingle the bell.

Like when I smell that cellophane
I'm like, "Ooh, new Cisco device.

Where is it?"

So it smells good and it is smooth to touch.

It is. You install these into racks -- you know,
I'm kind of tongue-in-cheek on saying that.

But this is the core of what our
networks use to connect devices.

One more piece that I want to add on
here, because it was on the title slide,

I didn't add it in my bullets, is
the concept of a broadcast domain.

A broadcast domain essentially means,

how far will a broadcast message
go before it's stopped, okay?

So a broadcast, broadcasts just
happen in the network world.

They're partially how computers work.

Like when this computer boots up, if it's
using DHCP, it doesn't have an IP address,

so it will send a broadcast message
saying, "Hello, anybody out there?

I don't have an address,"
an IP address, I should say,

"but if somebody could give
me one that would be great."

It's looking for a DHCP server.

Now, on a switch, just like a hub,
broadcasts will go to all ports.

That's one of our scalability concerns.

A hub you can only scale to
a few devices, I mean, 30,

40 devices you're starting
to really hit the max.

Switches you can get into
the hundreds of devices.

You know, 200, 300, 400 devices, but eventually
you're going to reach a point where you get

so many broadcasts because
everybody's kind of chitter-chattering

around that you're impacting your
computer and network performance.

So you start dividing it up
into more broadcast domains,

and we'll talk about how that's possible, but
I just wanted to throw that term out there

so it starts sticking in your mind.

All right, the last thing I want to do is
give you a day in the life of a switch.

Like if you want to see a switch's job day in,
day out, like how to make the doughnuts kind

of job, this is what it looks like.

We've got our network devices,
in this case five computers,

that are plugged into two different switches.

Now, the first thing I want to
do, I threw two switches up here

because I wanted to show you this cross-connect.

Now, you can connect switches together
on the Ethernet ports, no problem at all,

using just a normal crossover cable.

Or nowadays they have the auto-sensing ports
that allow it to detect a crossover straight

through and make the adjustments
accordingly, but that's one way to bridge them.

The challenge with Ethernet is that
you have a 100-meter limitation,

so once you exceed that,
now you're kind of stuck.

So a lot of people will start
going with things like fiber.

You know, a lot of times there'll be
buildings across the street from each other;

they need to run cables that are longer than
100 meters or just a really big building,

so they'll start putting fiber in here.

This is actually known as an SFP module.

It's a fiber-optic module
that you can slide in there.

You buy them separately.

I think SFP stands for "small form factor..."

pluggable.

I had to pause and look it up.

Everybody just says SFP.

For some reason the acronym
isn't used very often.

But this has a fiber-optic connection.

Now, on fiber you always have two,
I guess you could call them wires --

they're not really wires; they're
glass or plastic -- that are in there.

One is going to be a send and
one is going to be a receive.

So you kind of go click, click and plug it into
that, and then that fiber, depending on the kind

of fiber it is, you can, I mean you can
go 500 meters away, you could go miles or,

depending on what, where you
are in the world, kilometers,

depending on what currency you use for distance.

So you can span these things way
apart and connect them just the same.

So a lot of times on the switches you will see
these ports that are kind of dual-purpose ports

to where you can either, this is like a CAT
5 connection or CAT 6 copper connection,

and then below it will be an SFP where you
can plug in one of these fiber-optic modules.

Or you might see one, I think this is one of
them, I don't think these are dual purpose,

where you have four slots
where you can plug in SFPs.

Now you might say, "Why do
you have them changeable?"

Well, there's different kinds
of fiber you can get.

There's multi-mode fiber, which
is really easy to work with

and a lot cheaper to make because it's plastic.

And so multi-mode fiber has the ability of
being really easy to work with and really cheap,

but it doesn't go as far as single-mode fiber.

And single-mode fiber is always glass
that, you know, if you were to open this

up with a little razor, you would see
really, really thin glass that it's using

and a lot more difficult to work
with and a lot more expensive.

So based on what your needs are, you can buy
single-mode or multi-mode fiber interfaces

and then you just have to -- you've just
got to make sure they're compatible.

Make sure essentially the cable
type and the connector type.

There's different types of
connectors for fibers.

This is an SFP-style connector, but
they have big connectors, small --

you've just got to buy the
right cable for the job.

So that's how you connect them together.

Now, little fiber-optic lesson aside,

let's talk about the communication
and clear off all my gibberish.

When you first boot up these switches,
they have something known as a CAM table.

And the CAM table is essentially empty.

CAM stands for content accessible memory.

It's a place in memory where it stores stuff,
and in this case, the CAM table is going

to include our MAC addresses that it's going to
learn from our different devices on the network.

So when we first boot the
switch, it's completely empty.

So let's say, now we've seen our MAC addresses.

MAC addresses are 12 characters, so let's
just say this happy computer right here is 11;

1;11;11;11;11.

You'll see them written this way.

That's a common way to write a MAC address.

You'll also see them written this way.

This is a lot of times what you'll see in
the Cisco world: 2222:2222 or Microsoft a lot

of times, like if you open a
command prompt and do an IP config,

all -- Microsoft likes using dashes.

So, for instance, it'll show
you MAC addresses like this.

So I'll make this guy, let's just use
all three styles here on the screen.

33;33;33;33;33;33, so really
the style doesn't matter so much

as that there's 12 characters inside of it.

So every MAC address is 12 characters long.

So this guy, let's say the guy on the
left, the happy computer is talking

to the straight-faced computer in the middle.

So he's going to send a frame,
let's just say I ping...

let's see, I'm trying to think
of how far back I'll go here.

Okay. So let's just say IP
address-wise, this guy is 10.1.1.1;

this guy is 10.1.1.2 on an IP address.

So on this computer I type
in ping 10.1.1.2, right?

What's the first message
that's going to be sent?

An ARP -- address resolution protocol --

saying, "Okay, great, I see that you're pinging
10.1.1.2, but I've got to create a frame.

I've got to have source and destination IP
addresses, source and destination MAC address.

I don't have the MAC address for
10.1.1.2," so ARP is a broadcast message.

So it's going to say, "Hello, network.

Who is 10.1.1.2?"

So that hits the switch.

Now this is, both of these switches are
now considered one broadcast domain.

So it's going to receive that broadcast and
send it out to all ports that are active.

I mean, there's nothing plugged
in, so this guy gets the broadcast,

this guy, it shoots across to fiber.

These two get the broadcast
saying, "Who is 10.1.1.2?"

Now, this is the only one
that will respond to that one,

so the straight-faced computer
comes back and says "Oh, that's me."

So -- and I jumped a little ahead.

He goes, "Oh, that's me.

I am 10.1.1.2; this is my MAC address."

But let me take a step back.

As soon as this broadcast came into the
switch, I mean, if we were to grab that

and use Wire Shark and look at it, we
would say, "Okay, it's an ARP message.

It's trying to find a MAC address.

It's coming from the source
IP address of 10.1.1.1.

It's looking for the destination
IP address of 10.1.1.2.

It's coming from the source MAC address
of 11;11, you know, this guy right here.

And it's going to the destination
MAC address of, I don't know.

Now, you might say, "Well, wait a
second; it doesn't know this MAC address.

It has to have something in there."

Well, it absolutely does.

It's destination MAC address is FFFFFF.

Remember, MAC addresses are hexadecimal,
so A through F are valid characters.

And the very, very last possible MAC address
in all the scheme, FFFFFF, means a broadcast.

That's what that means.

So the switch receives it destined to
this MAC address that it will never learn

on any individual port and it says, "Okay,
well, that MAC address, it says go everywhere.

But I just learned something."

By seeing this broadcast message come into
the switch it goes, "Oh, oh, oh, wait a sec.

On port -- " let's just say this is port 0/2.

You know, this one up here was 0/1.

So it says "Okay, 0/2 is really 1111:1111:1111."

So now the switch has become
one MAC address smarter.

Now this guy comes back and says, "Oh,
that's me, let me send my ARP reply."

Here's his ARP message.

"I'm coming from the source IP
address going to this destination,

coming from this source MAC
address, going to this destination."

Let me just zoom in on that.

You know if I were to break that packet down,

it would say the destination
MAC address is the ones.

The source MAC address is the
twos and it will come into switch.

Now, two things happen here.

One is, let's say the switch is looking at,

let's just say that is port
0/12, or let's say 0/11.

I'm trying to be accurate looking at the switch.

Let's say 0/15.

So it's going to say, "Okay, I
just learned another MAC address

because I just saw a frame come in there.

It is 2222.

So I now know that that computer
is available on port 0/15."

Now, this is all in the CAM.

All in the memory of the switch, so if I
pull the plug on this switch at any time,

all of this goes away and it has to
relearn all of these MAC addresses.

So that's the first thing it does is it
learns that MAC address is on that port.

But now it's become smarter.

It goes, "Oh, it looks like you're
trying to go to the destination

of 111111, you know, all the ones.

I know where that guy is.

He's on port 0/2."

So now instead of sending the reply to all
ports, the switch is just going to say, "Okay,

let me switch you right over here down to
this device" and only those two get it.

And now point-to-point communication
can happen between those guys

without disturbing anybody else because
the switch has learned their MAC address.

Now, let's talk of real world.

Okay, a switch powers on.

How long is it going to take before
it learns all of the MAC addresses

on the network, or the ones that it needs?

I would say five to 10 seconds.

I know. Really, that fast?

Yeah. I mean hundreds of devices it can
learn in five to 10 seconds because all

of these guys are talking, and if you're
talking about powering on a switch, well,

as that switch is powering on, the devices
are chatting They need IP addresses.

They want to communicate.

So the switch is gaining knowledge essentially
right after it boots and is ready to do that.

It's ready to go, so it does not take
long for these guys to learn at all.

Also keep in mind these entries
have a life span of five minutes,

meaning if the device goes
quiet for five minutes --

and this is by default, you can change it -- the
switch will forget where that MAC address is.

So this guy goes quiet.

He disappears from the table.

Now, no big deal.

That just means the next time somebody has to
communicate with him, the switch will treat it

like a broadcast, because it's going
to say, "I don't know where 111111 is."

So it will send it everywhere
and then relearn that.

The reason it does that is
because MAC addresses can change.

Now, it's rare, really rare
for a MAC address to change.

But somebody could replace a network card.

You could be doing strange
things with virtualization.

We'll talk about that later on.

But I mean, there's times where the MAC address
can change, so this guarantees you that device

within five minutes will be able to
communicate or be learned or, you know,

that MAC address will be replaced
on that switch if it stops talking.

Welcome to the world of switching indeed.

That is how a switch functions.

That is its goal, is to bring
together all the devices

into a local area network and
allow them to communicate.

So what do we see, and what
do I want you to do with it?

You see the bullets on the screen.

Those are the major points that we hit.

First thing I would like you to do is grab a
piece of paper and in your own words write down,

"A hub is this" or if maybe you're not a
writer, you know, go to a friend, go to a spouse

and explain to them, "A hub is this.

This is the difference between
a hub and a switch."

Or, I do this sometimes: I'll be in Best Buy
or one of the places where you just go and...

waste money on nonsense, but you'll
be in Best Buy and talk to one

of the employees and see
their perspective on it.

I've got some very interesting results from
my surveys where you just go in and say, "Hey,

I'm looking for a switch, just a small one.

Where are those at?"

"Oh, they're over here.

Here's the box."

And then throw this kind of question out
there, just say, "Actually somebody told me

to get a hub, and someone
told me to get a switch,

and then someone else said
they're the same thing.

What do you say?

What's best?"

And see what they say.

That is one of the most fun -- now,
don't throw them under the bus.

Don't be like okay, "No,
this is really how it works."

But sometimes you'll get people who are like
ninjas, that are like -- they know exactly.

Other people will just say I don't know.

Other people will just make up the most creative
and unique answers that you've ever seen.

Sometimes they'll look at the
box and see if it says something.

But it's fun.

It's a fun survey to take.

The second thing I want you
to do is look at some devices

and identify how many collision domains,

how many broadcast domains
exist on each one of those.

Now, if these are all daisy-chained
together, if I take some cables and go clink,

clink and link all these together,
how many broadcast domains?

One. A broadcast will go
everywhere on those switches.

How many collision domains?

Well, start counting.

One, two, and that's the console
board, so don't count that one,

you know, start counting those up.

So if you're preparing for certification,
be ready to answer those kinds of questions.

You know, just based on these devices and
how they're connected, how many, you know,

you'll see some hubs mixed in with switches.

How many collision domains exist and all that.

Now, there's a question.

What if I take a hub and plug
it into that switch port?

Okay, now how many collision domains do I have?

You know, if I'm just talking about
that, that realm of it right now?

One. It's just one because it doesn't matter
how -- even though I'm plugging into a switch,

still only one device on that
port is able to send or receive.

We've gone back down to half duplex --
because we're in a hub -- at a time.

So again, drill yourself.

Go through systems like that.

And then finally, just understand
how the switch is doing what it does.

Look at that day in the life of a switch
and see how it learns some MAC addresses

and becomes just more efficient as it runs.

I hope this has been informative for you
and I'd like to thank you for viewing.