ipv6 private address

[Arjun Dabol]

Hi Experts,

I am trying to understand how to write a IPV6 Private address.

so for example I have an address : 134.23.33.22 and I want to write a IPV6 address equivalent to it , what that would be?

I know that IPV6 private address starts with "fc" so will it be like this ?

fc:134:23:33:22::/128 ??

plz suggest

[Michael Vincent]

You're thinking of "unique local" which can be used, although I've never seen it bothered with.

It's actually fc00::/7, so you'd use:

fc00::134:23:33:22

Although I'm not sure that's what you want.

Visually, that "maps" to the IPv4 address, but logically, it doesn't given that IPv6 addresses are represented in hex.  The logical mapping would be:

fc00::86:17:21:16

In any case, including IPv4 addresses in IPv6 is bad practice (unless it's specifically a 'mapped' address for translation) since you'll be porting any bad IPv4 practices into your new IPv6 scheme.

Cheers.

[Arjun Dabol]

Thanks Michael, Actually I need to give an IPV6 address just for lab purpose (there is some bug and workaround is to give any IPV6 address to the interface) , now since other end of the link is ipv4 address "134.23.33.22" I was planning to use same, just as IPV6 address on my side.

[James Leinweber]

For lab purposes you might be able to get away with the "documentation prefix" 2001:db8::/32.

To recap, global scope addresses currently start with 2, unique local with fc, link-local with fe, and multicast with ff. 

Is there a problem with getting a real IPv6 allocation from your ISP, even if you aren't routing it yet, and just using some subnets from that for your test lab?  That's what I did.

There are a few issues with the RFC-4193 style unique local addresses.

1. First, the "L" bit should currently be 1, so the prefix is actually "fd".
2. The next 40 bits are supposed to be chosen randomly.  Blithely using zeros is not random, and recreates the RFC-1918 IPv4 collision problems RFC-4193 is trying to prevent.
3. Many operating systems behave badly choosing source addresses if they see both global scope and unique local routes on a single interface.

The upshot is that most people are simply doing IPv6 with global scope addresses.

[Arjun Dabol]

thanks, Actually there is a reason for which I am doing this.

There is an issue if you try to bring BGP neighborship UP between an IOS and IOS XR router and both having IPV6 address family enabled.

You can see error "no global address configured" in sh bgp nei output. the workaround is to configure any IPV6 address on Loopback of IOS XR router.

so I wanted to use just any IPV6 address ..

i am planning some thing like ... "fc00:180:222:222:250::/128  where 180.222.222.250 is IOS loopback.