RE: ocean: do not boil

["Hesham Soliman (EAB)" <hesham.soliman@era.ericsson.se>]

  > >   > Why not?  How do we think that our attempts at v6->v4
  > >   > translation will be better than deploying an IPv4 NAT?
  > >
  > >=> I don't understand how you could convice someone
  > >that IPv6 is needed because of the existing problems with
  > >IPv4 (lack of addresses, NATs ....etc) but tell them that
  > >they need NATs anyway to deploy v6.
  > 
  > And, you think this will read differently if the "NAT" has
  > a "-PT" on the end?

=> It will read and deploy differently :)

  > >There is something
  > >broken in that logic. The difference between v4 NATs and
  > >NAT-PT is that, in time we will hopefully phase out NAT-PT
  > >or its use will be reduced significantly. You can't
  > >say the same thing about v4 NATs.
  > 
  > They phase out at the same time -- as soon as it isn't
  > necessary for any host within the site to communicate with
  > an IPv4-only host, the NAT(-PT) functionality won't be
  > used any more and can be removed.

=> They probably will but what is the driver for someone
building a network to deply both ? I can't see anything
getting simpler here.

  > 
  > There are actually some advantages to the IPv4 NAT approach:
  > 
  >          - IPv4 communication works inside and outside the
  >                  site.
  >          - Leverages existing infrastructure/expertise in
  >                  IPv4 networking.

=> I'm not addressing existing infrastructures, I was
referring to a fresh operator. In our DT's scope (cellular)
there will be a lot of those since the infrastructure is 
new.

  > The main disadvantages would seem to be:
  > 
  >          - Need to run dual-stack on all nodes that need to
  >                  access IPv4-only nodes or services.
  >          - Need to assign IPv4 addresses (probably requires a
  >                  DHCP server)

=> Need to manage IPv4 multiple NATs as well.
(I'm still talking in the context of our DT). 

  > 
  > I can see why those might be bigger issues for an all-new
  > installation, but do you really think they are blocking issues?

=> I don't know if they are blocking or not (an operator would 
need to answer that), but having to deploy IPv4 NATs certainly 
seems redundant and does not make anything simpler than deploying 
an IPv6 network with translators.

  > 
  > Would you agree that a current site that is using IPv4 behind a
  > NAT should deploy IPv6 side-by-side with the current IPv4 NAT
  > infrastructure?

=> If they have it already, yes, I agree, but that was not 
my point. Even if they already have IPv4 infrastructure
without NATs but forecast to need more IPv4 addresses in
future, I don't see the need to force them to deploy
v4 NATs.

Hesham