Roman4604 said:
The reason why there is no large scale momentum toward ipv6 is that currently there is no dire need.
With NAT (specifically PAT) becoming the norm due to security requirements (running on private addrs is one of the fundamental security design principals) and hosting software becoming virtualised (most web & mail hosting sw can support 1000s of domains on a single IP), there isn't a massive requirement for more addrs now.
I suppose in time this will change, but I think its some time away. No one is going to go through the expense & trouble if there isn't a pressing need.
This is a very unfortunate position to take, IMO. I seem to remember the scramble at the end of 1998 and 1999 to get ready for the Y2K problem. Five years before, when it would have been cheap to fix, no-one cared about it. The computing industry seems to wait until the last second before fixing problems.
According to
http://bgp.potaroo.net/ipv4 the current IPv4 reserve will be exhausted by 2016 (10 years away) and total exhaustion is predicted in 2022, by assuming linear growth. This is, even though a large part of the Eastern countries are already moving towards IPv6 and not consuming as many IPv4 addresses. A different curve being fitted on the data by CISCO (
http://www.cisco.com/web/about/ac123/ac147/archived_issues/ipj_8-3/ipv4.html) hints at exhaustion in September 2008 (See Figure 8)
This forum is striving towards universal broadband adoption in SA. If we assume a 10% penetration (of any type, broadband or dialup), with a population of 40m, it gives us 4m needed IP addresses. To allocate 4m IP addresses, you will need 62 Class B address ranges (and 62 more for 3G/GPRS/EDGE enabled cellphones & PDA's).
The current address allocation is also very unfair towards the different registrars. As an example, APNIC has been assigned 16 Class A ranges (268,435,424 addresses) while AfriNIC has been assigned 1 Class A (16,777,214 addresses). IANA (which until now also included Africa) has allocated to itself 40 Class A addresses (671,088,560 addresses). The unallocated pool is 64 Class A addresses (1,073,741,696 addresses) (see the links above).
NAT and PAT and CIDR is a stopgap measure to reduce the number of actual allocations, and does make sense from a security point of view. However, it does mean that no push technologies can be implemented, as the machines are completely unreachable and at the mercy of the NAT/PAT device. If it is unable to forward data, then no data will reach the device. Also, some NAT/PAT devices break some IPSec implementations, since it has to rewrite the packets before sending them out. It also makes P2P implementations (eg. Skype) very unreliable. Many protocols (eg. SIP) now has to implement workarounds just to allow the devices behind the NAT to be accessible. I believe you can get the same level of security with a restrictive firewall than a private range and a NAT/PAT device. Just make sure no connections are allowed to the internal LAN that didn't originate from there, then filter what's allowed to be originated from inside, and you're all set.
If you have two embedded devices running the same service on the same port, there would be no way to send data to either one or the other, the device will have to go out and find the data. In addition, your internet-connected PVR cannot easily subscribe to a software update or EPG multicast feed from eg. Multichoice, or your internet-enabled fridge subscribing to a multicast feed from PicknPay and Spar, determining the lowest prices for you before you order new items. Something like this would make total bandwidth consumption a lot lower, as with IPv6 you only receive the feeds you subscribe to.