I didn't mean to spark off another translation war ;-) I realise that there are downsides to every translation technique. I'm not trying to have the perfect solution, because I know that doesn't exist yet. Instead, I'd like to have a network segment that has the highest possible ration of IPv6 to IPv4, where all my users' application continue to work. The history here is as follows:
pre-2002: public IPv4 /24, and 'everything works' 2003: public IPv4 /24, and a public IPv4 /48. 'Everything works', but some apps only use IPv4 2011: NAT64/DNS64 test. Most stuff works over IPv6, and could do with a single IPv4 address, but some stuff breaks. <- not an option. The next step I'm trying to achieve is: 2014: <X> Can do with a single external IPv4 address, and and everything works. I thought it would be nice to try 464XLAT as <X> and get some practical experience with it. On 26 November 2013 10:47, Lorenzo Colitti <[email protected]> wrote: > On Tue, Nov 26, 2013 at 6:41 PM, Marco Sommani <[email protected]> > wrote: >> >> Well, I understand that there is value in moving towards an IPv6-only >> backbone, but, even in that case, I find it safer to provide IPv4 via >> DS-Lite rather than via 464xLAT. In my view, translating between different >> IP versions can only produce more problems. > > > Agreed. Actually, what you want is MAP-E. Unfortunately, in 3GPP networks, > the encapsulation used by MAP-E breaks DPI mechanisms that many operators > use for billing. -- Dick Visser System & Networking Engineer TERENA Secretariat Singel 468 D, 1017 AW Amsterdam The Netherlands
