| Forum Home | ||||
| Press F1 | ||||
| Thread ID: 47889 | 2004-08-08 08:18:00 | Better than IPv6?? | george12 (7) | Press F1 |
| Post ID | Timestamp | Content | User | ||
| 259503 | 2004-08-08 08:18:00 | Here's an alternative to IPv6. It's much easier to change to from IPv4, and still lets every person in the world have 157 IP addresses (if they were all shared out evenly). The addresses are 40-bit, written just like IPv4 except each bit between dots goes up to 1023. (eg, highest address is 1023.1023.1023.1023). Other than this change, keep it like IPv4. This would make converting routers and servers etc to the system. The addresses would technically only be a byte longer, so in terms of transmission time there is little difference. We could move up to IPv6 later, after this version, but for now giving 256 times the number of addresses would certainly help the shortage problem! Subnetting can be done just the same way, except starting with /40 instead of /32. For example, /40 1 address /39 2 addresses /38 4 addresses /37 8 addresses The greatest thing is that existing addresses can be used with this system, so address-wise it is backward compatible. 69.254.12.199 in IPv4 is the same as that in my version. Of course, 615.399.716.1002 cannot be used as an IPv4 address. So, many organizations converting would not have to renumber their network. Should they want to, however, they could take advantage of a new private range, 1.x.x.x - allowing for 1,073,741,824 addresses. Don't be too hard on me if this system is completely screwed up, as I don't exactly have a PHD in IP addressing (do they even exist?). George |
george12 (7) | ||
| 259504 | 2004-08-08 08:58:00 | You might have a good idea there, but what makes you think posting it in this forum will help it get anywhere. ?:| | kiki (762) | ||
| 259505 | 2004-08-08 11:34:00 | I always wondered why they didn't do that. Unfortunately, it achieves not much more than ease of use for us humans. Routers will still need to be changed - this is merely stalling big great leap. IP use is growing more exponentially than otherwise - therefore changing now would mean we have to change again shortyl afterwards. With IPv6, by the time they implement it, it'll last a LOT longer. Unfortunately, it will be as backwards compatible as IPv6. Nothing that we use now can migrate to this method without adding extra support - as you would need to to get IPv6. Also- How is 69.254.12.199 the same now as in your system? Now, in binary, it's: 1000101111111100001100011000111 With your new system it will be: 001000101001111111000000110000011000111 And because computers look at the binary as opposed to the octects, they wont get it, unless you could program them otherwise - but that takes alot of effort. |
Growly (6) | ||
| 259506 | 2004-08-08 11:39:00 | I think there are two main problems with the idea (aside from being a bit late :-)). The required changes to get everything to support your new scheme would still be roughly the same as IPv6. It probably would not be possible to get most routers etc to use it, so they would need to be replaced. Any any software that currently works with 32bit addresses would need to be updated. And for the actual implementation the 40bit size would be awkward because IPv4 (&6) breaks everything into 32bit words. It is unlikely you would be able to reclaim a spare 8bits from another header, so you would need to increase the header size by a full 32bits, leaving 24 wasted. It doesn't sound much easier than IPv6. Increasing the address size to only 40bits is going to be a short term solution. Why not solve it properly? (Although history is on your side, *cough* x86 *cough* windows). If you haven't already, check out the ipv6 specs for how they will handle backwards compatablilty (RFC 2460 & RFC 2373). It looks like they will do something similar to your idea, except the low 32bits of the ipv6 address are the same as the ipv4 address. IPv6 is more than just a move to 128bit addresses, it also adds new features to the protocol. |
bmason (508) | ||
| 259507 | 2004-08-08 11:50:00 | It's hard enough remembering a string of 12 digits :( But I like the idea . We shouldn't run out of IP addys! The way I see it 64 . 6 . 378 . 123 . 4 64 = New Zealand 6 = Area Code 378 = Local Exchange 123 . 4 = Actual Number Phone number for example . Would that just be SO easy? Then again, what about networks? BUT there is more than 255 countries (I think) . So, have the numbers go up to 500 . 500 . 500 . 500 . 500 Just a theory . -=Tom=- |
mr_rix (5375) | ||
| 259508 | 2004-08-08 12:07:00 | Given the now idealistic nature of the conversation, I'd like to revise my theory: I want IP addresses to be more like computer numbers. Every computer in the world numbered, decimally, from 1 onwards. Then we can start numbering the people instead of giving them names.\ Example conversation. Person#_1025162: So, friend, do you wanna play an Internet Game of Doom 7? Person#_1025982: Sure, I don't see why not. What is your Unique Internetwork System Identifier Code? Person#_1025162: Twelve. *Person#_1025982 connects to an internet game of Doom 7 at UISIC 12. |
Growly (6) | ||
| 259509 | 2004-08-10 11:48:00 | What I meant when I said that 69.123.21.200 would be the same in my version is that the only conversion needed would be adding 2 zeros to each um *whats the word for bit between the dots?* chunk in binary. The number is still the same. You could still type in the number, and it would work. Try typing in 69.123.21.200 as an IPv6 address. I just don't think we need to change absoloutly everything like IPv6 does. I realise there are other improvements, but think that a simpler, more similar version would solve the not-enough-addresses problem quickly, and ISPs could afford to easily give people like me more than a stupid single dynamic IP address that changes every time my crappy USB adsl modem decides to die. As to the exponential growth thing and 32-bit word thing, how about 64-bit numbers? Like this: 65535.65535.65535.65535 = 1111111111111111.1111111111111111.1111111111111111 .1111111111111111 I find that a lot easier to remember than something like: fe80::2c0:4fff:fecf:9973 (real address) And people could use IPs just they way they did, but with a lot more combinations. IP addresses would be just as easy to remember for a long time until they got into 4 digits (xxxx.xxxx.xxxx.xxxx). But, until there was need for more than 996,005,996,001 addresses, IP addresses could stay xxx.xxx.xxx.xxx! This 64-bit address could support up to: 18,446,744,073,709,551,616 addresses, over 2,635,249,153 per person on earth today! So, in the future where we all live in space and dwell on all 9 planets plus 11 others we haven't found yet (not my life beliefs), every person in the world can still have over 131 million IP addresses. That's never going to run out. I think that because current IP addresses can still work, transition would be much easier. The room-size massive routers in the core of the internet could convert first, while people convert at their own pace and in the meanwhile use simple conversion programs that add 00000000 to each "chunk" between the zeros. Oh well, it's never gonna happen, time to accept IPv6. |
george12 (7) | ||
| 259510 | 2004-08-10 12:43:00 | As I see it, both IPv6 and IPv4 use 8 bits (0-255) for each block of the address. Your proposal uses 10 bits and therefore instantly requires massive worldwide hardware upgrades? As the IP data is "serial" in nature, its likely that 8 bits will continue for compatability, only the number of blocks will increase, which perhaps is accommodated more by software? |
godfather (25) | ||
| 1 | |||||