[SOLVED] IPv6 private address spaces

dyscoria · 2011-02-12 11:43:48

I'm trying to figure out the ipv6 equivalent to the ipv4 private address spaces:

10.0.0.0 to 10.255.255.255
172.16.0.0 to 172.31.255.255
192.168.0.0 to 192.168.255.255

The ipv6 private address space is described as "fc00::/7". However, my understanding of CIDR is pretty limited. My first guess for the range that this represents is:

fc00:0:0:0:0:0:0:0 to fc00:ffff:ffff:ffff:ffff:ffff:ffff:ffff

But it seems to me like this is actually fc00::/16

Last edited by dyscoria (2011-02-12 12:08:21)

wudu · 2011-02-12 11:53:45

Hi /7 means 7 bits from left are subnetmask

-> fc00:0:0:0:0:0:0:0 to fdff:ffff:ffff:ffff:ffff:ffff:ffff:ffff

fc00 -> 11111100 00000000
fdff -> 11111101 11111111

hope I'm right

dyscoria · 2011-02-12 11:59:15

Thanks wudu!

I've also found this article which seems to confirm what you said:

http://blog.ipexpert.com/2010/08/02/ipv … explained/

I've been trying to get my head around CIDR/hexadecimal/subnetmask/bits...my brain seems to be functioning a little slow today

dyscoria · 2011-02-12 12:07:51

So, link-local address, which are described as FC80::/10 are:

FE80:0:0:0:0:0:0:0 to FEFF:0:0:0:0:0:0:0

I think I understand this now. Awesome.

I really do find that the address range is so much more readable than the CIDR notation!

Last edited by dyscoria (2011-02-12 12:11:09)

cedeel · 2011-02-14 11:38:20

dyscoria wrote:

So, link-local address, which are described as FC80::/10 are:
FE80:0:0:0:0:0:0:0 to FEFF:0:0:0:0:0:0:0

No.

fc80::/10

1111 1100 1000 0000 ::

for which /10 is

1111 1100 1000 0000 ::
to
1111 1100 1011 1111 ::

same as fc80:: to fcbf:ffff:ffff:ffff:ffff:ffff:ffff:ffff

fukawi2 · 2011-02-15 03:05:06

Note that IPv6 Private Addresses ("Site-Local") have been depreciated by RFC3879:
http://www.rfc-editor.org/rfc/rfc3879.txt

ghen · 2011-02-15 09:02:59

See RFC 4193 for Unique Local Addresses (fd00::/8).

It is recommended to select a random /48 range from there – so not just fd00:0000:0000:: or anything that is simple / looks cool, to somewhat guarantee global uniqueness and thus to avoid collisions when companies/sites/networks merge later (10/8 has caused many such headaches in IPv4).

fukawi2 · 2011-02-15 11:44:54

ghen wrote:

...to somewhat guarantee global uniqueness...

That is just an epitome of failness IMHO. I can't believe it's an RFC

ghen · 2011-02-15 15:24:55

fukawi2 wrote:

That is just an epitome of failness IMHO. I can't believe it's an RFC

These addresses are explicitly not globally routable, so global uniqueness is not required.
It is only -potentially- an issue at the time two ULA networks are merged. But 40 bits of randomness reduces the chance of collisions practically to zero. The RFC is explicit about the risk though.

If global uniqueness is a requirement, just use public IPv6 address space, that's why/how IPv6 was designed. The ULA range is intended primarily for sites with no Internet connectivity.

Last edited by ghen (2011-02-15 15:30:12)

fukawi2 · 2011-02-15 22:00:31

ghen wrote:

The ULA range is intended primarily for sites with no Internet connectivity.

In which case IPv4 would be sufficient
I haven't heard of any private networks exceeding the addresses/subnets provided by 10/8

ghen · 2011-02-15 22:53:22

fukawi2 wrote:

In which case IPv4 would be sufficient
I haven't heard of any private networks exceeding the addresses/subnets provided by 10/8

There certainly are. eg. Comcast ran out of 10/8 for addressing their cable modems. Some large corporations do run out too, certainly if they didn't plan their numbering too well in advance. An IPv4 /8 is still "just" 16M addresses.

But this topic was about IPv6, right? :-)

fukawi2 · 2011-02-16 04:43:37

ghen wrote:

There certainly are. eg. Comcast ran out of 10/8 for addressing their cable modems.

That's not a private network with no internet connectivity... The Comcast issue is what IPv6 is here to solve

dyscoria · 2011-02-17 11:01:30

cedeel wrote:

dyscoria wrote:
So, link-local address, which are described as FC80::/10 are:
FE80:0:0:0:0:0:0:0 to FEFF:0:0:0:0:0:0:0
No.
fc80::/10
1111 1100 1000 0000 ::
for which /10 is
1111 1100 1000 0000 ::
to
1111 1100 1011 1111 ::
same as fc80:: to fcbf:ffff:ffff:ffff:ffff:ffff:ffff:ffff

Thanks!

I finally get it now (for real this time). I don't think my brain is made for CIDR notation

Last edited by dyscoria (2011-02-17 11:01:52)

Arch Linux

#1 2011-02-12 11:43:48

[SOLVED] IPv6 private address spaces

#2 2011-02-12 11:53:45

Re: [SOLVED] IPv6 private address spaces

#3 2011-02-12 11:59:15

Re: [SOLVED] IPv6 private address spaces

#4 2011-02-12 12:07:51

Re: [SOLVED] IPv6 private address spaces

#5 2011-02-14 11:38:20

Re: [SOLVED] IPv6 private address spaces

#6 2011-02-15 03:05:06

Re: [SOLVED] IPv6 private address spaces

#7 2011-02-15 09:02:59

Re: [SOLVED] IPv6 private address spaces

#8 2011-02-15 11:44:54

Re: [SOLVED] IPv6 private address spaces

#9 2011-02-15 15:24:55

Re: [SOLVED] IPv6 private address spaces

#10 2011-02-15 22:00:31

Re: [SOLVED] IPv6 private address spaces

#11 2011-02-15 22:53:22

Re: [SOLVED] IPv6 private address spaces

#12 2011-02-16 04:43:37

Re: [SOLVED] IPv6 private address spaces

#13 2011-02-17 11:01:30

Re: [SOLVED] IPv6 private address spaces

Board footer