Well, think of it like that, once all hosts support IPv10 (as all hosts now support IPv6) network engineers will cope up with the new header, because the destination will be from a different version, Old firewall configuration will be the same but new configuration will be addred because new IPv6 only devices will be on place (this is their normal job).
>> Q: In your IPv10 world, what happens when an IPv4 only host (and by that, I >> mean a host that cannot talk anything but plain vanilla IPv4) tries to talk >> to an IPv6 only host ? This host must be updated, no reason not to accept the updates. >> Q: In your IPv10 world, what happens when one of your new format packets >> hits a router/firewall/load balancer/other network equipment that doesn't >> understand it ? These devices must be updated, that's why the word ALL answer all your questions. Khaled Omar -----Original Message----- From: Int-area <[email protected]> On Behalf Of Simon Hobson Sent: Monday, September 28, 2020 10:22 AM To: IPv6 Operations <[email protected]> Cc: int-area <[email protected]> Subject: Re: [Int-area] [v6ops] Still need to know what has changed.... Re: IPv10 draft (was Re: FW: v6ops - New Meeting Session Request for IETF 109 - IPv10) Khaled Omar <[email protected]> wrote: > Ok, I have no energy to keep repeating, I'm sorry, read the full draft please > again. And therein lies a big part of your problem. Constantly repeating something which is incorrect does not make it correct - well not in our world anyway ! You can say "it's not a problem" as many times as you like, that does not make it so. People have tried to explain to you in many ways that upgrading hosts that are easy to upgrade is not the problem - that's mostly solved. Firewall rules/ACLs/etc do not stay the same - they must all be updated to cope with the new combinations of addresses that can be employed. Routers everywhere must be updated - believe it or not, IPv4 routers are usually hardware tied to the format of an IPv4 packet - they simply will not understand any other packet. And by requiring a different packet format end-end you are enforcing that your packet format cannot be used (reliably) until 100% of the internet has been upgraded. And as already said, pretty well all the changes needed to support your IPv10 are much the same as needed to support IPv6. But, with IPv6 once you get there then you've reached the end-game (IPv6) rather than having reached just a milestone on the way to IPv6. And there are migration options to handle "bits" (whether that's hosts, bits of the internet, services, whatever) that still can't manage IPv6. So, some specific questions you've been asked but refused to answer. I say refused, because saying "just read the draft again, you don't understand" is refusing to answer legitimate questions about what is in your draft. The questions asked show that people have looked at and understood your draft - they can see reasons why it cannot work, but you are failing to address those issues. Q: In your IPv10 world, what happens when an IPv4 only host (and by that, I mean a host that cannot talk anything but plain vanilla IPv4) tries to talk to an IPv6 only host ? A: It fails. If you believe otherwise, please explain how it works - without requiring the host to be IPv10 capable. In reality (in the absence of network based migration techniques), it'll do an A lookup in DNS, get no result, and just fail. It won't know what to do with an AAAA record. Bonus answer: Explain how hosts that are no longer supported, or have hardware restrictions (limited RAM/ROM space), are owned/used by people who have no idea what a "firmware upgrade" is and have even less idea why they'd want to do one, etc get upgrades ? Q: In your IPv10 world, what happens when one of your new format packets hits a router/firewall/load balancer/other network equipment that doesn't understand it ? A: The packet gets dropped<period>. If it isn't an IPv4 packet, then an IPv4 device won't handle it - by definition, your IPv10 packets are not IPv4 packets. Your IPv10 host cannot communicate along that path. Thus it must fall back to trying either IPv4 or IPv6 - so it might as well have just used IPv4 or IPv6 in the first place. Again, if you believe my answer to be wrong, then explain in detail why - not just "you don't understand, read the draft again". Simon _______________________________________________ Int-area mailing list [email protected] https://www.ietf.org/mailman/listinfo/int-area _______________________________________________ Int-area mailing list [email protected] https://www.ietf.org/mailman/listinfo/int-area
