Got it. NAT is more problematic than duplicate addresses in a private address
space.
I will try to rewrite the document with IPv6 in mind. I may have some concrete
points to add to prove the need for the virtual address space and ip address
reclassification.
One such point is about virtual neighborship through the virtual address space.
For example virtual EBGP local neighborship and virtual EBGP global
neighborship. I think this could be implemented with ULA secure Neighbor
discovery and global IPv6 addresses. Through RR clusters dynamic path
recalculations can be performed to resolve the congestion problems.However the
virtual address space is still needed from my perspective as a local and global
controller.
Thanks & Regards,Vineet Deshpande
On Tuesday, 2 October 2018, 6:28:17 pm GMT+5:30, Mikael Abrahamsson
<[email protected]> wrote:
On Tue, 2 Oct 2018, vineet deshpande wrote:
> Is IPv6 striving to eliminate NAT completely or is it only to avoid NAT usage
> for conserving address space ?
NAT was invented to handle IPv4 exhaustion. With IPv6 there is no address
shortage, so no need for NAT to avoid address limitations.
The basic premise of the Internet is that it's interconnected networks
with unique addresses. If you don't want something to talk to/from the
Internet, then you firewall it or don't announce that address space to the
Internet. There are some other variants, like ULA that amounts to the same
thing.
If your load balancer re-writes addresses as part of its normal operation,
that's up to you. There is no need to call the addresses before and after
anything special, they're only special in IPv4 possibly due to address
number limitations so they can be re-used in several locations. No need
for this in IPv6 unless you absolutely want to. I would advise not to.
--
Mikael Abrahamsson email: [email protected]
_______________________________________________
Int-area mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/int-area
