re: Poll Shows Divisions Over IPv6IPv6 is coming. The poll seems to indicate this (56% say it will be widely used on either 5 or 10 years). Furthermore lots seem to say it will start in AsiaPac.
Fine.
But should this protocol arrive via dual stack or via its mandatory use in various system standards?
For example, what is the real benefit of mandating IPv6 in the 3GPP/IMS specs for release 5? Wouldn't it be better to support both IPv4 and IPv6 and so initial rollout could be made on a more stable platform and then we move over to IPv6 when the stacks are bugfree(ish) and the 3G market actually needs the address space.
re: Poll Shows Divisions Over IPv6Not to denigrate the Unstrung IPv6 poll, but a search of publications from years past would reveal polls that foretold the inevitable widespread adoption of OSI protocols to replace IP, CMIP management to replace SNMP, token ring to replace Ethernet, broadband ISDN to replace CATV, etc. etc. etc.
Participants in such polls are often too close to the fray to offer much objective insight.
Question: Why adopt IPv6? Conventional Wisdom: Because it yields more address space.
Q: Why not just use IPv4 with network address translation to reclaim address space? CW: Because translation is a kludge that breaks things.
Q: So the transition from IPv4 to IPv6 will happen instantly, as a flash cut? CW: Of course not, the transition will take months or years.
Q. Then how will IPv4 and IPv6 systems communicate? CW: Through dual stacks, tunneling, and translation.
Q. But wait a second... You say IPv4 with NAT is unacceptable because translation is a kludge, yet you say IPv4 to IPv6 transition requires translation. Are you contradicting yourself? CW: Stop asking questions, and go check out how the latest poll said IPv6 adoption was inevitable.
Maybe Unstrung could run a poll about 3G WCDMA deployment, so we can prove to ourselves that it will enjoy inevitable widespread adoption, too!
You are right to question IPv6, but your main argument that IPv6-to-IPv4 tunneling is a kludge as much as NAT is too simplified! The difference is that IPv6-to-IPv4 at least has the potential of being a TEMPORARY solution until IPv6 gets widespread enough and gradually that kludge/overhead will be reduced. Staying with NAT/IPv4 give you no hope of improvement!
However, one thing that should worry you is the bandwidth overhead and additional processing required for the increased IPv6 addresses and headers. Of course, that is a vendors' blessing since it would require HW upgrades :-) However, if the need is strong enough that is a penalty customer will have to pay
However, there are other benefits of IPv6, I don't know them all, but I understand improved operation (i.e. fewer of the unintended drawbacks of NAT) through firewalls is one (Don't ask me why/how).
I'd love to hear from some IPv6 diehards on what other nifty things there are (that would make sense to end-users/operators!)
re: Poll Shows Divisions Over IPv6x-Eri >> I'd love to hear from some IPv6 diehards on what other nifty things there are (that would make sense to end-users/operators!)
Doesn't sound like there is anyone out there that will support IPv6!!
And what about my original point - If you are going to need dual stack etc. why do we mandate IPv6 only for 3GPP/IMS specifications? Why not start with IPv4 and then transistion later on??
P.S. And if anyone thinks IPv6 is "complete" go and read the huge list of 3GPP dependencies on IETF at http://www.3gpp.org/TB/Other/I...
Fine.
But should this protocol arrive via dual stack or via its mandatory use in various system standards?
For example, what is the real benefit of mandating IPv6 in the 3GPP/IMS specs for release 5? Wouldn't it be better to support both IPv4 and IPv6 and so initial rollout could be made on a more stable platform and then we move over to IPv6 when the stacks are bugfree(ish) and the 3G market actually needs the address space.