What's the Freakin point of a Link-Local address?

volfkhat

Hola,

an IPv4 address can be Over-simplified into 2 types:
~Public Address type.
~Private Address type.

Public Address is routable Externally,
Private Address is routable Internally.
Nothing Shocking here.


Now compare that to IPv6:
~Public Address equivalent ---> Global Unicast Address, (2001-3FFF):: .
~Private Address equivalent ---> Unique Local Address, (FC00::)

Okay, makes sense to me.
So....

What's the Freakin point of a Link-Local address?
It doesn't route Internally, it doesn't route Externally, it doesn't Route at ALL.
As far as i can tell... it's the IPv6 version of APIPA (you know, 169.255.x.x).
Who cares! APIPA is virtually useless. You wouldn't assign any local hosts to it in IPv4.
So WHY is it now required in IPv6??

I just don't get it...

xnx

Fallback for DHCP I think

volfkhat

volfkhat wrote: »

Now compare that to IPv6:
~Public Address equivalent ---> Global Unicast Address, (2001-3FFF):: .
~Private Address equivalent ---> Unique Local Address, (FC00::)

Unless i have Backwards?
With IPv6, Global Unicast Address is used to route both External & Internal traffic?

Thus, IPv6 has essentially rendered "private" ip addresses obsolete,

?

volfkhat

xnx wrote: »

Fallback for DHCP I think

I don't follow.

Why is it Required for IPv6 then?
(it's Not required in IPv4)

?

networker050184

Here is one example. Say you had a giant network with 10,000+ p2p links between routers. You can take the time and assign a /126 on all these links or you can just enable IPv6 and let the routers communicate with each other via link local. Save a little time pushing the config and documenting the usage, and to top it off you minimize the attack surface of your router as you say they are nonroutable! There are some cons in this scenario as well, but it's just an example.

volfkhat

networker050184 wrote: »

Here is one example. Say you had a giant network with 10,000+ p2p links between routers. You can take the time and assign a /126 on all these links or you can just enable IPv6 and let the routers communicate with each other via link local. Save a little time pushing the config and documenting the usage, and to top it off you minimize the attack surface of your router as you say they are nonroutable! There are some cons in this scenario as well, but it's just an example.

I "think" i see the point you are trying to make.
And I understand that the "current" behavior of IPv6 is to let the Protocol effortlessly assign link-local addresses to each host:
FE80::EUI-64

But my Counter-Argument is simply: Why doesn't IPv6 use the Unique-Local instead?
FC00::EUI-64

You get the same end-result, Plus everything would be internally routeable.
?

mikeybinec

for arp

volfkhat

mikeybinec wrote: »

for arp

??
There is No ARP in IPv6.


But assuming you meant NDP...
what does that have to do with why Link-Local addresses are required (instead of Unique-Local addresses, for example)?

Additionally, ARP/NDP operate at Layer 2...

But I'm asking about Layer 3.

Mow

Great article by Fish:

Understanding IPv6: Link-Local 'Magic' - Network Computing

volfkhat

Mow wrote: »

Great article by Fish: Understanding IPv6: Link-Local 'Magic' - Network Computing

Yes, but it doesn't seem to answer the Question at hand.

Allow me to Rephrase:
a Unique-Local address (FF00: can do EVERYTHING that a Link-Local address (FE80: can do.
However, a Unique-Local address can do One additional thing: Route Internally.
(In my opinion... pretty big Advantage)

So with that in mind...
Why didn't the Architects of IPv6 make the Unique-Local address the requirement?

Instead,
why did they decide that Link-Local address needed to be the requirement

?

volfkhat

Okay, i have arrived at this conclusion:

In IPv4:
Your host-NIC can have only One ip address.
It can be a PUBLIC, PRIVATE, or APIPA address.... but it can only be 'one' of them (at any point in time).

In IPv6:
That same host-NIC can have TWO ip addresses (at the same time).
One address is Routable: Global-Unicast or Unique-Local,
the other Address is Non-Routable: Link-Local.

So... Why is the Link-Local address now MANDATORY in IPv6?
Honestly, I still do Not know.
lol

but I think the best explanation is to simply "pretend" that IPv4 ALSO allowed Two ip addresses.
And that this 2nd address, the "Link-Local", was simply 'hidden' from our view.
:]

Jon_Cisco

It sounds like this question is making you mad. You are probably just approaching it wrong.

IPv6 is not suppose to map to IPv4 it is suppose to replace it and when they decided to make changes this is one they felt was useful.

james43026

It's quite simple actually. A IPv6 link local address is exactly the same as an APIPA address in the IPv4 world. They both serve the same exact function.

volfkhat

james43026 wrote: »

It's quite simple actually. A IPv6 link local address is exactly the same as an APIPA address in the IPv4 world. They both serve the same exact function.

Absolutely true!
Except... your device is NOT required to have an APIPA address in IPv4 :]

Jon_Cisco wrote: »

IPv6 is not suppose to map to IPv4 it is suppose to replace it and when they decided to make changes this is one they felt was useful.

I see where you are going from.
But i guess my question to you is: Why did they feel the need to make this specific change? What was the added "usefulness"(as you put it)?

/shrug
(and personally, i would argue that IPv4 and IPv6 actually do MAP to each other)

Oh well... Fun stuff, nonethless :]

james43026

volfkhat wrote: »

Absolutely true!
Except... your device is NOT required to have an APIPA address in IPv4 :]




I see where you are going from.
But i guess my question to you is: Why did they feel the need to make this specific change? What was the added "usefulness"(as you put it)?

/shrug
(and personally, i would argue that IPv4 and IPv6 actually do MAP to each other)

Oh well... Fun stuff, nonethless :]

In the Internet Protocol Version 6 (IPv6), the address block fe80::/10 has been reserved for link-local unicast addressing.^[3] The actual link local addresses are assigned with the prefix fe80::/64.^{[7][note 2]} They may be assigned by automatic (stateless) or stateful (e.g. manual) mechanisms.
Unlike IPv4, IPv6 requires a link-local address to be assigned to every network interface on which the IPv6 protocol is enabled, even when one or more routable addresses are also assigned.^[8] Consequently, IPv6 hosts usually have more than one IPv6 addressassigned to each of their IPv6-enabled network interfaces. The link-local address is required for IPv6 sublayer operations of the Neighbor Discovery Protocol, as well as for some other IPv6-based protocols, like DHCPv6.
In IPv6, stateless address autoconfiguration is performed as a component of the Neighbor Discovery Protocol (NDP),^[9] as specified in RFC 4862. The address is formed from its routing prefix and the MAC address of the interface.
IPv6 introduced additional means of assigning addresses to host interfaces. Through NDP routing prefix advertisements, a router or a dedicated server host may announce configuration information to all link-attached interfaces which causes additional IP address assignment on the receiving interfaces for local or global routing purposes. This process is sometimes also considered stateless, as the prefix server does not receive or log any individual assignments to hosts. Uniqueness is guaranteed automatically by the address selection methodology (MAC-address based according to RFC 4862, and/or randomized according to RFC 4941) in combination with the duplicate address detection algorithm.

This was taken from Wikipedia. Hope this helps point you in the right direction.

mikeybinec

volfkhat wrote: »

??
There is No ARP in IPv6.


But assuming you meant NDP...
what does that have to do with why Link-Local addresses are required (instead of Unique-Local addresses, for example)?

Additionally, ARP/NDP operate at Layer 2...

But I'm asking about Layer 3.

yeah, that was a bad response on my part. It does put your mac address in the host part and as you correctly said, ndp works from there+-

theodoxa

Link Local addresses are widely used in IPv6. IPv6 Routing protocols for example will use the Link Local Address of the next hop. You could just use Link Local addresses on all interfaces connecting devices. If they had done this with IPv4, it would have conserved a lot of IP Addresses as IPv4 uses up 4 addresses (1 for Network ID, 2 Usable, and 1 for Subnet Broadcast) on every point-to-point link. IPv6 would have simply assigned a Link Local address to each end of the Link.

james43026

volfkhat wrote: »

Absolutely true!
Except... your device is NOT required to have an APIPA address in IPv4 :]




I see where you are going from.
But i guess my question to you is: Why did they feel the need to make this specific change? What was the added "usefulness"(as you put it)?

/shrug
(and personally, i would argue that IPv4 and IPv6 actually do MAP to each other)

Oh well... Fun stuff, nonethless :]

Actually IPv4 and IPv6 can not map to one another without assistance, IE NAT64. Solutions like IPv6 over Ipv4 tunneling only works to allow IPv6 traffic to traverse a IPv4 network on it's way to another IPv6 network, a IPv6 only host cannot directly communicate with an IPv4 only host without the assistance of some kind of translation between the protocols.

volfkhat

Hi James,

Poor word choice on my part; i should not have said "map".

I just meant that, Initially, i could Not wrap my mind around ipv6.
But Now i can See many SIMILARITIES :]

Ipv4 has 4 octects.
Ipv6 has 8.

Ipv4 has APIPA.
Ipv6 has link-local.

:]