All language subtitles for 34 - Routing - IPv6 Concepts-eng

>> The year was 1998 when I was
teaching a Microsoft IIS class,

a phenomenal product responsible for about 90
percent of the security flaws in Windows server

but nonetheless, I remember a student
asking me something like, "Jeremy,

when do you think we'll be on IPv6?"

And I said, "2003," I don't know why.

It's some of those things you just remember.

Totally missed it, but that was my prediction.

And here we are many, many years later, nearly
a decade, oh, actually more than a decade later,

and we're still waiting to fully migrate
to IPv6 but what I can say, it's happening.

I have seen things changing in
these last few years that I'm like,

"Okay, this is now becoming a reality."

A matter of fact, Cisco even adding IPv6 to the
CCNA program is a huge statement right there.

So that's what this nugget's all about.

Kind of the big picture, what is IPv6 all about?

What is the new addresses look like?

What are the new kinds of addresses?

But before we get in to all the addressing
details, we need to define the why.

Why are we going to IPv6?

Well, the first thing I say is, "Yes, Virginia,
there realty is an IP address shortage."

And the reason I say there is, is because we've
almost become numb to hearing the statement

"We're out of IP addresses" and
yet, miraculously they all appear

from somewhere whenever somebody needs them.

But it is true, the government has
given out all of the IP addresses so,

they're now all in the hands
of private entities.

They're not officially allocated by any means.

There are entities, government
entities, universities,

businesses that are literally sitting
on hundreds of thousands if not millions

of IP addresses that are just sitting there
unused because they were given to them back

in the early days of the internet
before anyone knew it would get so big.

And they're kind of like "Well,
why we should give it back?

Even if we're not using them, we
might as well hang on to them."

So, we just have a poor allocation and even

if we could somehow fairly allocate all the
IP addresses, we would still have a shortage

because we have all kinds
of new devices on the rise.

Everything can have an IP address.

Microwave ovens, stoves, refrigerators, I
mean, and then I'm not talking like the future.

I'm like this is now.

You can go buy a stove.

You can buy a microwave that has an IP
address so that you could run diagnostics

on it remotely, you can get remote control
and set timers and all that kind of stuff.

I mean our television has moved
to IP based systems, you know.

The old, you know, bunny ears antennas,
those are all kind of fading away in place

of everything being connected
to this one network.

Now, you can get into car-- I mean, so the
point is we've got all kinds of devices

that are [inaudible] and the problem is NAT
which is the savior of the internet, go NAT,

is now seen as a hindrance to
innovation, because if you think about it

from a business perspective,
businesses have been like "Okay, IPv6.

We're going to upgrade but why are we going
to-- are we going to make more money at this?"

The answers like "Well, no."

There is some cool new features of
IPv6, but it's not anything that would--

a business would be like, "Oh, yeah, let's take
on the expense to change our entire organization

to this protocol because we will
definitely see return on investment."

No, it's not a protocol like that.

It's just, you know, we're
reaching the point where we have

to have more IP addresses and
that's the main motivation.

So, rather than seeing this light switch effect
where it's like, you know, it's like Y2K, right?

One day you wake up and poof, the internet is
now IPv6 and my computer doesn't work anymore.

What we're going to see is a slow migration.

I mean it's already begun right?

If you-- this is the world if you're
having trouble recognizing that.

What's happening is key stakeholders on
the internet, take like Google, take Cisco,

you know, all these different entities all

around the world are slowly
enabling their servers for IPv6.

There was a not too long ago, I think it
was a year or two ago, they had IPv6 Day,

it was like a big news announcement to
where all of these big companies have said,

"Now we have our websites available
on the IPv6 version of the internet."

So, so you're seeing the slow train migration
to IPv6, however, the more and more we're going

to start hitting a wall to where, you know,
IPv4, it's only going to be able to stretch

so far and you're going to kind
of see this wall to where now,

it's going to be like IPv6 internet
connections are going to become more common

and equipment is going to be shift-enabled
for IPv6 by default rather than IPv4.

And it's just going to be kind of the
slow fading effect to where it's not going

to be a light switch, it's just going
to be like, you know, 30 years from now,

somebody would be like "Oh,
yeah, I remember IPv4.

Whatever happened to that?"

You know, it just kind of slowly faded away.

Now, there are future features
that are, you know,

some small ones now being developed
constantly, for instance, IPsec everywhere.

IPsec is part of the IPv6 suite of protocols.

Meaning it supports it natively just
like, you know, IPv4 has TCP and UDP.

IPsec is a protocol of security.

It's used a lot of times for VPN connections.

So, you can actually have encryption turned
on by default for all of your traffic so,

everything is secured at the box.

Mobility becomes easier to where you can
have like a cellphone or a device roaming

between different routers,
different access points.

Simpler header makes our
equipment far more efficient.

We actually have a slide
on that in just a moment.

So the best way I can put it is there's
not going to be a smoking gun like "Oh,

look at that feature, let's all move to IPv6."

It's not-- people aren't going to move
to IPv6 necessarily because they want to,

although I'm sure there's plenty who do.

It's going to be one of those we have to.

You know, at this point, we need this.

We can't get anymore IP addresses
from our service provider

of their routes so, let's make the move.

We've got to cut over.

So, let's look at the address system itself.

Address size has moved from 32-bit
addressing in IPv4 to a 128-bit addressing.

Now, you guys are all subnetting
masters by this point, right?

So, you know that, you know, when we take
an IPv4 address and have its four octet,

each one of those are eight bits,
zeros and ones that make it up.

Well, with IPv6, what they have moved to,
actually let me look down here real quick.

They have moved to eight octets of IP addresses.

And now we've moved to a
hexadecimal to where each one

of these represents 16 hexadecimal bits.

So, if you take 16 plus 16 plus 16,

you start adding all those up,
that's what leads us to 128.

Moving to this scheme has
provided this mini IP addresses.

I have yet to meet person who can
actually say that number although it is one

of the few numbers that is bigger than
the national debt, but I thought that to--

that this website is, I thought, just
a great one that shows all of it.

It says, "How many IP addresses
does it support," you know,

and they kind of show "Here's the number."

How do you say it-- you know, there's
no way without resorting to math.

And so, I thought this guy-- this was one

of the most amazing facts if
you can really understand it.

He say, "Here it is in real world
terms, it's big as in grains

of sand don't even enter into how big IPv6 is.

We have to go to an atomic level."

So look at this, it says, "So
we could assign an IPv6 address

to every atom on the surface of the earth."

Okay. Wow, okay, they put that into
perspective, that's a lot of addresses,

and still have enough addresses left
to do another hundred plus earths.

So, I know some people are
saying, "Well, you know,

people didn't think we would run out with IPv4."

That's true.

People were like "Wow, this
provides billions of addresses,

why would we ever need more than that?"

so that, you know, IPv4 was
seemed as the unfathomable end.

But honestly, I mean read that again.

I mean every atom on the surface of the earth
and still have enough for a hundred plus earths?

Okay, I know I don't have the
foresight of some but I would say,

"I can't ever see running out
of IP addresses with IPv6.

If we change to IPv7 in the future, it's
probably going to be for a reason other

than the fact that we're out of IP addresses."

This is just my thoughts but, take that from a
guy who thought would be here in 2003, right?

So, here is technically how you say it
and this is just to impress your friends

if you want to memorize that list.

Its 340, I can-- I don't even know how to
pronounce that, undecillion, 282 decillion,

none-- I mean, these are numbers that
I just-- I don't have experience with.

So that's a lot of addresses
that we have available to us.

So, first things first, why
did we use hexadecimal?

How come we did that instead of using decimal?

Hexadecimal is friendly for big numbers, meaning
it shortens the amount you would have to write.

If you took a 128-bit address and, you
know, took it in our current IPv4 format,

you would have to add 16 octets total.

You know, if each one were eight bits a piece
as it is currently, 16 octets, you know,

that is just not a friendly number to
remember or to write or anything like that.

It's even worse if you were
to write the thing in binary.

But hexadecimal uses numbers zero through nine

and then it adds on A through
F on there as well.

So, each one of these digits in the
IPv6 address actually is represented

by four bits of information.

So, you know, 0000 is still zero.

0001 is still one.

But the way it works is as
you move up in binary,

you know, this is still three, still four.

And you move up in binary and hit nine, the
next one up instead of moving to 10 moves

to the letter A, and then B, and then C.

So, technically, you can represent
16 digits or zero through 15

with every single one of the hexadecimal digits.

That allows us to have an IP address
that has a lot more scalability,

like it represents a lot more devices or a lot
more addresses and not make it so big to write.

But still, I mean still looking at this,
it's huge compared to our IPv4 address.

So, the first thing that they were concerned

about when they developed the standard
is, how do you make it shorter.

Well, the first one, rule number one is you can
eliminate groups of consecutive zeros inside

of the IPv6 address and you'll see when we
start looking at some of the addressing.

Typically, there are a few groups of
consecutive zero that you can take out.

Now, with this rule, and so
let's look at what it does first.

If we have an address like this, I would see
"Okay, it looks like all zeros, all zeros,

all zeros, there's three octet straight."

It's not really called an octet anymore but
I'm stuck in that world so let me call it that.

Three octet straight of zeros, that
can be represented by a double colon.

Now, key point on this.

You can only use the double colon one
time and only one time in this address.

So, for example let's just
imagine that we had, you know,

a couple more sets of zeros here at the end.

I couldn't have colon, colon and then,
you know, this would all be scribbled out

and we have colon, colon because the
device wouldn't know how many sets

of zeros go in each spot.

I mean computers are pretty smart if they
know an IPv6 address is eight octets,

or whatever we call those now, long, and it
says, "Okay, well, I see one, two, three, four,

five here and there's a double
colon, the computer goes, "Oh, well,

if it's eight octets long and I only see five
then there must be three octets of all zeros."

You know, it kind of fills in the blank.

Does that make sense?

But that's why we can only
use the double colon one time.

That can be a big shortener for us.

The second thing, rule number two is
we are allowed to drop leading zeros,

not trailing zeros but leading zeros.

So for example, let me focus
in on this one right here.

0050, right?

That can be reduced down here to just 50.

The leading zeros can be dropped off.

I mean we can continue on, you can
see AB4 was 0AB4 and that's all I got.

It's all I got in that address.

But, so essentially, I mean, still a
long address compared to our IPv4 days,

but compared to this, it's significantly,
at least half, shorter than what it was.

In addition to increasing the sheer
quantity of IP addresses that are available,

the standards bodies that created
IPv6 also wanted to simplify it.

They wanted to simplify the header.

The current IPv4 header is smaller simply

because the addresses aren't
as big as the IPv6 addresses.

However, it's got a lot of
complexity inside of it.

Now, it doesn't look a like a lot but
really when a router is processing thousands

and thousands and tens of thousands
of these things, every single second

that it's sitting there doing business,

that's a lot of processing time that's wasted
looking at these headers inside of here.

So, with IPv6, I mean it's bigger and--

just because the address size but really
when you look at how many fields it has,

it's much simpler for the router to process,
much simpler to look at and be like, "Okay,

got it, let's move, move forward."

So our routers can be more efficient.

They accomplish that using
this next header field.

Next header allows you to use something called
extension headers to where, for instance,

other headers can be put in behind this that
add characteristics like maybe you need some

of these fields, or you need some custom fields.

Maybe you've got your own propriety
protocol that you want to create.

Well, the cool thing is this.

The next header can point to those
extensions and can process those, you know,

based on whatever kind of
routing platform you want.

So, the next header field was probably one of
the most important things that was added in here

to eliminate the complexity out
of the IPv6 protocol itself.

Okay, now let's talk about the
kinds of communications and the kind

of addresses that we see with IPv6.

First off, the familiar.

We have Unicast addresses which represent
one-to-one communication from one device

to another, that's just the same as
it was in the IPv4 address suite.

Multicast, same thing as it was before,

it is we now can have one
message go to a group of devices.

So, you know, if I've got two computers
here that are part of the Multicast group,

my one message can go straight to them.

So one packet going to two devices and
not bother the third, far more efficient

that sending one packet to you, one packet
to you, especially useful for things

like imaging computers where I just want to
send one stream to a group of computers or,

for instance, internet broadcasting to
where I have a radio station, or movie,

or television channel online where a bunch of
people are coming, I just want people to be able

to tune in to it rather than sending
individual streams because if I rely solely

on Unicast traffic, it hits
the wall on scalability.

You know, the more devices that jump
on and start tapping into the stream,

the more streams the server has to send.

Whereas Multicast allows you to have kind
of like radio broadcasting, send one message

and all of these people tune in to it.

Now, a new one that has entered
the fleet is Anycast.

Anycast. That is one-to-closest.

Kind of a cool thing, we've had all kinds of
methods of load balancing come out with IPv4

but nothing that's integrated
into the protocol itself.

Well, with IPv6 addresses, you
can assign an Anycast address

which in a nutshell means
I give the same IP address

to multiple devices maybe around the world.

Maybe I've got-- maybe I'm
running amazon.com, right?

And that we have servers that
are in the United States.

We've got servers in Australia.

We've got servers in Japan.

I mean the Amazon has servers everywhere.

Well, with IPv6, you can just
give them an Anycast address.

I mean there's more to it than this but in a
nutshell, they all have the same Anycast address

and when somebody goes to Amazon,
it finds the closest server to them

which naturally load balances and gives you
the best response time 'cause distance-wise,

you're always finding the
closest one to you, very cool.

Now, does any one notice one
kind of communication missing

from this list that does exist in IPv4?

Got it? Broadcast.

Broadcast are gone in IPv6.

They have been completely
replaced with Multicast.

Now, I mean, it's like "Well,
how do we even work?

How do we do that?"

Well, multi-- they have Multicast addresses
that essentially do the same thing as broadcast.

So, when I say broadcast is gone,
just that word is gone, right?

The word broadcast is gone but
there is Multicast addresses

that reach everybody on the subnet.

So, essentially, just the functionality
of broadcast has now been grouped

in to the Multicast world as well.

Okay, so those are how we can communicate.

Now, let's look at the kinds of
IP addresses that we can have.

The first thing to get used to in
IPv6 is having multiple IP addresses.

In IPv4, you could do that but
most people did it on, you know,

advanced server configurations or,
you know, strange things like that.

It wasn't normal by any means.

But in IPv6, it's going to happen all the time.

First kind of address that you can
have is a link-local scope address.

What this is for is local communication like
in your same layer two switch infrastructure.

It's very similar-- anyone ever seen
the Microsoft addresses 169.254.

something, right?

So Microsoft setup a system to where
if your computer was setup for DHCP

and your network either did not have a DHCP
server or maybe the DHCP server was down,

the computers would make
up their own IP address.

They would say, "Okay, well,
I'm going to generate,

I'm just going to makeup an IP
address 169.254.50.91," you know,

and what it does is send a broadcast out
saying, "Hey, does anyone have this?"

And the response is normally, "Nope."

And it says, "Okay, well,
this would be my address."

The reason that they created it is it
allows devices to communicate maybe

on like a small office or home office network
where maybe the admin doesn't have the knowledge

or the equipment to setup a DHCP server.

It's a good idea but nobody really
used it because you can't get

to the internet with these addresses.

They are completely, completely blocked.

Well, in IPv6, they said, "Well, let's create an
IP address that is used just for communication

within the same switch infrastructure, and let's
have the computers make it up by themselves."

Now, as I'm saying this, it doesn't have

to always be this way, but
for the most part, it is.

And so, the way it works is the computers
will automatically generate an IP address

with FE80 at the beginning.

Like that, that's a very beginning of it.

You'll a lot of times see
people say, FE80colon, colon/64.

I said 64 but my pen just
amazingly wrote 16, /64.

Because what you'll find is a lot of people
represent the first 64 bits of the address

as the network and the last 64
bits as the host of that network.

That's-- you'll see that commonly shown in IPv6.

Now, why and how and, you know,
Jeremy can you fill that in?

Well, sure.

So the last-- so let's take an address, right?

Let me start here.

Take an address and-- I just need more room.

Actually, let me go to this direction.

When you've got FE80colon,
colon/-- why do I do that?

Something in my brain is stuck on 16.

FE8-- maybe 'cause this is 16 bits.

FE80colon, colon/64, what that
really means is the first 64 bits

of this address represent
the network and essentially,

up to a certain point, it's all colons.

So we could think of that as FE80 and using
our shortening methods 0, 0, 0, right?

So this represents the first four
octets of a link-local address.

That would be considered the network.

It's 64 bits.

16 plus 16 that's 32; 48,
64, so there's our 64 bits.

Now, the last 64 bits is left for the host.

The computer itself can autoconfigure itself.

Some people call this state list configuration.

Autoconfigure itself with its own host ID.

There's many ways to do this but let me say one

of the most common is actually
referred to as EUI-64.

I know. Right about now,
you're like "Whoa, whoa!

This started simple.

We're getting deep quick."

Don't worry, it's not too bad.

EUI-64 says, "I am going to use
the MAC address as my host ID.

Now, wait a second, MAC address.

MAC address if I remember
right was like this, right?

00-00-00, I'm using dashes like Microsoft does.

It was essentially 12 characters
which is good but short, right?

Because we need 16 characters if we're looking
at 64 bits, it's 16 hexadecimal characters

of four bits, we're short a few characters.

So, the way that UI standard was
written, UI-64 standard, it says,

"I'm going to take my MAC address."

Let's say my MAC address was
11, 22, 33, 44, 55, 66, right?

So, that's my MAC address.

I'm going to use that but I'm going
to split it right in the middle.

I don't know why a surgeon comes to
my mind when I think of this like,

slice it open right in the middle.

And I'm going to squeeze in FFFE
right in the middle of that.

So, the host ID ends up being 1122, 33FF, FFEE--
wait a second, wait a second, what am I dong?

This is all going together without 16 bits.

Hang on, I'm writing that wrong.

I know that some of you are like "Man,
I thought I was getting this but no."

FFFE, there we go, and then
hang on, where is my MAC.

I'm-- see I'm writing on the
screen, I can't see whatever--

I'm left-handed so I'm right behind myself.

So, 5566, right?

So, does that make sense?

It took this MAC address and which
is by the way a 48-bit address

and it said, "Well, I need 64-bit address."

So what I'm going to do is squeeze the
16 bits FFFE right in the middle of it

and that will now generate the host address.

So, you will see a lot of
auto-generated machines that say, "Okay,

my address is FE80colon, colon" and then you see
this EUI-64 essentially MAC address with FFFE

in the middle of it, generated after it.

So, if a device needs to
communicate to something

on its local segment-- let's relate it routers.

Like for instance, let's
say you've got two routers

where there is a switch in the middle of them.

Like CDP messages might sent
with IPv6 link-local address.

OSPF neighbors might be formed.

Those hello messages might be formed
using the link-local address which says,

"You are a local neighbor
to me as you should be."

So, a lot of that local communication
essentially all

within that same layer two
infrastructure is all reduced now

down to the link-local scope address.

Now, let's move in to the unique
and site-local scope address.

So, again, local subnet only, right?

Unique and site-local address is the
direct equivalent to our private addresses.

So, in IPv4, we have the 10 network, 172.16, all
those, those private addresses that we all know

in lab, they created a unique
or site-local scope address

which mirrors that for environments.

Now, I would say, this address type has been the
address type of most controversy as in it's come

in and out and back in to the standard as people
have argued around the functionality of this.

Essentially, if there're enough
global addresses, meaning--

and this by the way is our
equivalent of public addresses.

If there's enough public addresses around
the world that everybody can have one

or a thousand or, you know, every atom

of the earth can have one then why
do we need private addresses anymore?

And the answer still goes, well,
because that's what we're used to.

I mean that's what most organizations are used
to having this and there are other uses for it

but I will tell you, you can
definitely get by with this.

Now, I will tell you it's weird.

It is absolutely weird to say, "Every device
on my network has a public IP address."

In our IPv4 language, people
would be like, "Whoa!

Security vulnerability, hello,
what are you doing?"

Waste of IP addresses, all these things come to
mind, but the fact is that's where we're going.

Every device, I mean I'm kind
of bleeding down to this.

Unique or site-local scope, you
don't have to have one of those.

You could just go with your link-local
and a global scope address or internet

or public IP address on every
one of your devices.

Now, does that mean that our
fire walls come into play?

Yes, it does.

This does not mean that every device

on your network is immediately
fully accessible from the internet.

No. We have to have security in
mind but that's totally doable.

We've been doing that for years
with our current IPv4 addresses.

So, that's the kind of mindset.

Those are the three things of
addresses that we can have.

Let's dig a little bit deeper
into the global addresses,

and to see how they're going to workout.

First off, how are they assigned?

The powers that be which happens to be the
IANA, Internet Assigned Number Authority,

the same people who handed
out IPv4 addresses have said,

"We're not just going to
hand these out on a whim.

You know, obviously we want to have order
and structure and how these are given

out to the general world as it stands."

So, they have said, "We are
only going to handout addresses

with their high level bits set to 001."

Essentially, that will encompass the
entire world as we know it today.

So all of these addresses that are being handed
out all start with typically the number two,

and let me describe what that is.

Remember each one of these
are hexadecimal, right?

Each one of these digits are actually
represented by four binary numbers.

Remember hexadecimal can be one through
nine, or A through F, right, as values.

So that's essentially 16
values, zero through 15.

So they're represented by these four binary
numbers which can give you 16 different values.

So they've said, okay the first
three bits are going to be 001.

Now, that is typically represented by two.

Now, does it always have to be two?

No. We could have a one here and it could be a
three, right, as that first digit but if you--

I mean go on Google and type in global addresses
IPv6 and you'll see every single example

that you find will have a two in front of it
just because that's what everybody is using,

there is enough addresses to get there filled.

That is the beginning of what
you can call the global prefix.

The best way to understand this,
I think, is to really think

about how these are being handed out, right?

So we have the IANA sitting over here.

They've got this bajillion, you know,
undecillion addresses at their disposal

and they're saying, "Okay, service
provider A, please come here."

Service provider A says, "Yes, here I am.

I am AT&T."

They said, "I am going to give you the global
prefix and then they will typically dole

out a 48-bit global prefix for that customer."

Now, it doesn't have to 48 bits, it could
be less but it definitely can't be more.

So, let me just show you what that looks like.

So the IANA might say to AT&T, "Here
is yours, 2000, 1111, 2222, right?

That is yours."

That is actually a 48-bit
hexadecimal address right there.

No, it's not a full address,
it's just a beginning of one.

Now, how do I know that?

Well remember, every hexadecimal
digit is four bits in length.

So, every-- now, I keep calling
these octets just

because they haven't really come
up with a good word for them.

There's actually a standards committee that is--

they're trying to figure out,
what are we going to call these?

In IPv4, we call them an octet but
octet represents eight, like eight bits.

So, what do we call these?

I mean there're all kinds of suggestions.

But for now, I'm just going to
call them all octets, right?

So each octet of an IPv6
address is actually 16 bits.

So, we say, okay 16, 16, 16, so
that's 48 bits of the address.

Now, what service provider A will
do is say, "Thank you very much.

I now have 2000, 1111, 2222 and
now I have all of my customers."

So we have customer A, customer B, customer
C, and so on and so forth that we're going

to start handing them out to and they will take
the last 16 bits of that, so they'll say, "Okay,

well, I'm going to start with 0000."

That one goes to customer;
oh, I put two customer As.

That one goes to customer B. 2000, 1111, 2222,
0001, that's going to up here to customer A.

So then-- what we're doing is adding an
octet onto here, adding a thing onto here

that will represent a 16-bit subnet.

You notice the subnet ID is 64 bits-- 64
minus whatever we have as our global prefix.

Now, it doesn't have to be 48 bits.

It could be 32.

It could be 16.

It could be-- I mean there's different
things it could be but I will say,

huge amounts of these chunks are being handed
out and most commonly they are all 40 bits

in length, giving the service
provider the flexibility to come

up with their own little subnets.

So now, they give the 64-bit
network ID over to customer A,

and now that the customer has the
ability to start taking the last 64 bits

and identifying their specific
interface, or breaking it up even more.

I mean we could have a subnet to where the
service provider gives the customer a /56.

Because the /64 doesn't really give
the customer any flexibility, right?

Essentially that runs it right out
to where the network typically ends

and the interface ID begins, you know, the
interface ID generated from the MAC address,

or it could be typed in, or anything like that.

So doing a /56, what's that do?

56 is, again, let's take our 16
bits, we've got 16, 32, 48 bits.

Now, if we add another 16,
that's 64 but 56 is not far away,

56 happens to be eight bits less, right?

I'm writing really small there.

Eight bits less than that.

So, we kind of cut this right in half so
we would say, if we give a customer a /56,

maybe AT&T says, "Okay, 2000, 1111, 2222, 00"--
you know, scribble that out and do /56 goes

to you customer A. And now customer
A can sit there and say, "Okay, well,

I can now subnet that where
I've got 0001, 0002, 00--

" you know, they can create their own
little glob of 256 subnets using that.

Hang on. I have that feeling right
now, it's getting little muddy.

So, I mean, let me-- can I draw
that again and just show you

on a nice big white board what this looks like?

IANA, right?

So they are peered with all these
different service providers.

So we've got service provider A, service
provider B, and you know, down and down we go.

So we've got AT&T, we've got
Sprint, we've Cox Communications,

we've got all of these different service
providers are out there, so let me just use Cox

because that's one that I wrote up.

So, IANA says, "Cox, I'm going to give you
2000,1111,1112 as your /48 global prefix, right?

That's a big chunk that we're assigning to you."

So, Cox can then go to their
customers and Cox has customers A, B,

C and D. Let me just put some letters on them.

And Cox says, "Okay, we're
going to subnet that further.

So, we have been giving this but we know that
the network is represented by the first 64 bits.

So we have between 48 and 60-- " I'm
showing you IPv6 subnetting here.

Isn't this crazy?

We're just learning about the
addresses and here we are, right?

So, I've got between these that I can
use as flexibility for my addresses.

So, watch this.

They cans say, 2000 like this, colon and let's
just do 00/56 goes to customer A. That means

up to 64, the customer still
has two digits left, right?

So, customer still has two digits left, right?

So, customer A can now take that and say,
"Well, thank you we've got, you know,

five networks in our organization so
this one will be 2000 ah, ha, ha, ha."

I'm going to have to get used to that
on writing this big old addresses.

0000 will be our first subnet /64, right there.

This one will be dada, dada, dada,
0001/64, are you catching this?

You see now how this goes?

0002/64 and then, you know, on here
the last 64 bits will be for all

of the individual computers
that are sitting on that subnet.

They can, you know, this last
one dada, dada, dada 0003.

Essentially, they have now eight bits right here
that they could generate 256 individual subnets.

Now, this company only has five but hey,
we've got more addresses than there are atoms

on the planet, why not go ahead and take
this for the future of your organization?

It's no loss to us that you can now use
these and subnet them however you want.

Wow! Tell me, wow.

Isn't-- are you starting to see
this, how this allocation works?

And in my opinion, the subnetting
gets a little easier.

I mean if you-- now, I know we're all coming
in with some IPv4 addressing skill so,

looking at these values, you're
like "Okay, I got all that."

So, I know we're coming with some
previous knowledge but, you know,

for the amount of time it typically takes
to learn IPv4 subnetting, to see this,

it's like "Okay, that's not too bad.

If I know each one of these digits
represent four bits and, you know,

these represents how many bits are in the
subnet mask then I can easily find out.

Okay, there's where my line is
based on however I have this."

So, let's go back.

Okay, now the rest of the
bullets kind of fall in place.

The subnet ID is comprised of the bits
left over after the global running prefix.

And I should put on there dot, dot,
dot, and before the interface ID, right?

Because you've got the global
prefix, it's, you know,

IANA is handing out to some
service provider a /48.

Well, between that and the 64 bits of the
interface ID, you have the subnet ID, you know,

the 64 minus N bits that's squeezed in there.

So the primary address expected
to comprise the IPv6 internet are

from the 2001colon, colon16 subnet.

It's just based on what has been handed
out so far, who knows if that will stay

but I would say, for the short-term,
you'll probably see a lot of Internet2

or IPv6 internet addresses all
starting with 2001colon, colon.

You know, that's the first 16.

So, you know, you could have /32--
I'm just giving some examples.

Some /32 subnets, given to the providers
or could be a /48 subnet is given

to a service provider like AT&T,
and then they start subnetting that,

you know, with their subnet ID.

So, think of this, /32, I should put global
prefix is assigned to providers, right?

That's probably a little
better than saying subnet.

And then they create all of these little
subnets that they can give to their customer.

This is just an example, you know,
/48 subnets can be handed out as well.

Either way, do you see the millions and millions
of addresses that these will provide for anybody

who wants to jump on the IPv6 internet?

I mean there're more addresses
than atoms in the world.

We can dole them out without
really thinking much about them.

All right, I actually forgot I
had a whole slide for this but,

I've already described the
link-local addresses already.

Remember these are the ones
that are assigned automatically.

I just described them all in that first slide.

They're very similar to the
169.254 begin with FE80.

That's what it looks like in binary,
followed by 54 bits of zeros.

Last, 64 bits are the 48-bit MAC address,
this is the EY-64 with FFFE squeezed

in the middle so I'm giving an example here.

And so, these are all what I was showing before.

But this gives you a little more
"not squished in the middle" view

of how the link-local addresses are generated.

Wow, a big change, right?

That was actually deeper than I thought we would
get, but we covered some really good ground

in that nugget, looking at why we would--
why we would need to upgrade to IPv6?

You know, what's the whole point?

And again, the point is there's no real major
benefit other than the fact that we're going

to have to get there because the
IPv4 address space is running out.

So people will be somewhat forced into
a transition eventually by equipment,

by the standards that are being
developed by however body else is talking.

We started looking at the
IPv6 addressing format.

First off, what the addresses look like and
then we got deep quick getting into the headers

and then the address types looking at the
communication like Unicast, Multicast,

Anycast and then the link-local,
site-local, global address types knowing

that each device can have multiples.

And then finally, we went on that in-depth
exploration, like I said, a little more depth

than I thought we would get but really
seeing how the global addresses work

and how they're going to be allocated
by service providers in the future.

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