Author Topic: Forums ipv6 test  (Read 5906 times)

0 Members and 1 Guest are viewing this topic.

Offline chrisgbk

  • Inactive Staff
  • Veteran
  • *****
  • Posts: 1739
Re: Forums ipv6 test
« Reply #20 on: February 25, 2008, 02:00:19 am »
In particular the one misguided individal who believed that IPv6 was synonymous was static addresses for every device

That would be awesome.

You misunderstand what was meant by said person;...

Humm... that would be rather stupid.
Isn't the MAC address static? But you can't search the internet via MAC addr can you...

A MAC address is only quasi-unique; not guaranteed to be unique. The original address assigned to a device never changes, but it's a trivial task to spoof a different MAC address, and many routers in particular support this functionality.

Offline FliesLikeABrick

  • Administrator
  • Flamebow Warrior
  • *****
  • Posts: 6144
    • Ultimate 13 Soldat
Re: Forums ipv6 test
« Reply #21 on: February 25, 2008, 12:01:52 pm »
In particular the one misguided individal who believed that IPv6 was synonymous was static addresses for every device

That would be awesome.

You misunderstand what was meant by said person;...

Humm... that would be rather stupid.
Isn't the MAC address static? But you can't search the internet via MAC addr can you...

A MAC address is only quasi-unique; not guaranteed to be unique. The original address assigned to a device never changes, but it's a trivial task to spoof a different MAC address, and many routers in particular support this functionality.

Plus MAC addresses are only visible on your local subnet.  Your MAC address is only broadcasted as far as your router.  MAC addresses are used to identify machines on links, so outside your local network it would then be the MAC of your cable modem and whatever devices are on that network, and then the next network would have the MACs of that network, etc, etc.

So there is no way for, for example, me, to know what your MAC address is without being in your house.

edit: Chris, you haven't answered me yet :P  The question (about whether an AAAA record should be added) extends to anyone else who is familiar with IPv6 as well.

Offline chrisgbk

  • Inactive Staff
  • Veteran
  • *****
  • Posts: 1739
Re: Forums ipv6 test
« Reply #22 on: February 25, 2008, 12:17:00 pm »
Plus MAC addresses are only visible on your local subnet.  Your MAC address is only broadcasted as far as your router.  MAC addresses are used to identify machines on links, so outside your local network it would then be the MAC of your cable modem and whatever devices are on that network, and then the next network would have the MACs of that network, etc, etc.

So there is no way for, for example, me, to know what your MAC address is without being in your house.

edit: Chris, you haven't answered me yet :P The question (about whether an AAAA record should be added) extends to anyone else who is familiar with IPv6 as well.

A better question would be, is there any reason not to add one? I don't see any reason not to, as you have IPv6 connectivity, and adopting the future tends to be a good thing. Might as well support the few people who have access to IPv6 now.

PS: sorry for killing the quote train, but it was getting too long :P

Offline FliesLikeABrick

  • Administrator
  • Flamebow Warrior
  • *****
  • Posts: 6144
    • Ultimate 13 Soldat
Re: Forums ipv6 test
« Reply #23 on: February 25, 2008, 02:46:38 pm »
The main (only?) problem with people serving content on a dual-stacked machine under the same hostname is that some users have their applications and operating systems configured in such a way that their application/OS/etc thinks it has IPv6 connectivity, but it really doesn't.  This will result in, say, Firefox having an A and AAAA record, and attempting to connect to the IPv6 address when it really doesn't have a functinal ipv6 connection.

In this situation, the site would either take 10 minutes to load after every image and stylesheet request times out and falls back to ipv4, never loads at all, or at least times out once (then the application falls back to ipv4 for all subsquent requests).

Mind you this happens to only people who have actively fucked up their configuration, or otherwise have very bad default configurations.

I suppose it is worth a test at first, like if I scheduled a 1 hour time where the forums would have A and AAAA records, and then wait to see what kinds of problems people had (if any).  If no problems we could try it again for a longer period of time, etc etc.

Thoughts?

Offline chrisgbk

  • Inactive Staff
  • Veteran
  • *****
  • Posts: 1739
Re: Forums ipv6 test
« Reply #24 on: February 25, 2008, 09:21:34 pm »
The main (only?) problem with people serving content on a dual-stacked machine under the same hostname is that some users have their applications and operating systems configured in such a way that their application/OS/etc thinks it has IPv6 connectivity, but it really doesn't. This will result in, say, Firefox having an A and AAAA record, and attempting to connect to the IPv6 address when it really doesn't have a functinal ipv6 connection.

In this situation, the site would either take 10 minutes to load after every image and stylesheet request times out and falls back to ipv4, never loads at all, or at least times out once (then the application falls back to ipv4 for all subsquent requests).

Mind you this happens to only people who have actively fecked up their configuration, or otherwise have very bad default configurations.

I suppose it is worth a test at first, like if I scheduled a 1 hour time where the forums would have A and AAAA records, and then wait to see what kinds of problems people had (if any). If no problems we could try it again for a longer period of time, etc etc.

Thoughts?

I'd say go for it.

Offline a-4-year-old

  • Veteran
  • *****
  • Posts: 1918
Re: Forums ipv6 test
« Reply #25 on: February 25, 2008, 09:28:59 pm »
is there any benefit to you? It won't go any faster, it won't make breakfast. you don't really gain anything from using it except maybe you might feel good about being an early adopter. Thats about it.
If we hit the bullseye the rest of the dominoes will fall like a house of cards. Checkmate. -Zapp Brannigan

Offline FliesLikeABrick

  • Administrator
  • Flamebow Warrior
  • *****
  • Posts: 6144
    • Ultimate 13 Soldat
Re: Forums ipv6 test
« Reply #26 on: February 25, 2008, 09:48:48 pm »
is there any benefit to you? It won't go any faster, it won't make breakfast. you don't really gain anything from using it except maybe you might feel good about being an early adopter. Thats about it.

Being an early adopter is always good, plus in this case it means a tiny bit more content available on ipv6.  When it comes to moving to new technologies, early adoption is necessary in order to build any kind of momentum.  I know this is a tiny bit... but I do what I can. 

My involvement with Soldat has been a learning experience from the beginning.  Primarily, this is an extension to that experience.

Offline bja888

  • Flagrunner
  • ****
  • Posts: 745
  • Working
    • Bja888.com
Re: Forums ipv6 test
« Reply #27 on: February 26, 2008, 05:00:41 am »
The main (only?) problem with people serving content on a dual-stacked machine under the same hostname is that some users have their applications and operating systems configured in such a way that their application/OS/etc thinks it has IPv6 connectivity, but it really doesn't. This will result in, say, Firefox having an A and AAAA record, and attempting to connect to the IPv6 address when it really doesn't have a functinal ipv6 connection.

In this situation, the site would either take 10 minutes to load after every image and stylesheet request times out and falls back to ipv4, never loads at all, or at least times out once (then the application falls back to ipv4 for all subsquent requests).

Mind you this happens to only people who have actively fecked up their configuration, or otherwise have very bad default configurations.

I suppose it is worth a test at first, like if I scheduled a 1 hour time where the forums would have A and AAAA records, and then wait to see what kinds of problems people had (if any). If no problems we could try it again for a longer period of time, etc etc.

Thoughts?

I'd say go for it.

I'd say, run it my MM.

Soldat isn't exactly known for attracting the best and fastest computers you know. I think you might run the risk of excluding some of the community. Not a very large percent of course but it could always be the the tipping point that sends one guy postal.

Offline FliesLikeABrick

  • Administrator
  • Flamebow Warrior
  • *****
  • Posts: 6144
    • Ultimate 13 Soldat
Re: Forums ipv6 test
« Reply #28 on: February 26, 2008, 09:43:47 am »
I wouldn't do it if nobody could get here, that's what I was saying.

Also, what does this have to do with "the best and fastest computers" ?

Offline O.R.I.O.N.

  • Veteran
  • *****
  • Posts: 1710
  • Thunk-a-dunk.
    • Diseased Productions
Re: Forums ipv6 test
« Reply #29 on: February 26, 2008, 12:53:58 pm »
Changing the system could just as easily give us a headache as it could save time and effort. All in all, it'd be up to you. I don't really see how any website would need the sheer volume of address space that IPv6 provides, but if it works, hell, why not use it?
To sum up my point: We had a multipage debate about toilet padding. (Putting TP in the water so you don't get splashed.)
And we still don't know if dead guys can keep a stiffy.

Offline FliesLikeABrick

  • Administrator
  • Flamebow Warrior
  • *****
  • Posts: 6144
    • Ultimate 13 Soldat
Re: Forums ipv6 test
« Reply #30 on: February 26, 2008, 02:05:43 pm »
Changing the system could just as easily give us a headache as it could save time and effort. All in all, it'd be up to you. I don't really see how any website would need the sheer volume of address space that IPv6 provides, but if it works, hell, why not use it?

Think about a home network with 5-10 computers.  As of right now they all (typically) share one IP address on the ISPs network.  This sucks, and it makes it a lot harder to get a lot of things work.  The reason this is done is because the ipv4 address space isn't big enough for each home (or business) connection to get a *range* of addresses to use to assign to its computers.  IPv6 addresses this.

When it comes to web or game servers, it is right now hard and sometimes very expensive to get a number of IP addresses.  For example, the soldat server, wizard.u13.net, has 2 IP addresses: one for the Apache instance, and one for the lighttpd installation.  To get that second address, I had to contact my host and explain to them why I needed it.  I'm lucky, my host doesn't make me pay extra for more addresses, I just need to explain why I need it.  Most webhosts require you to pay more for IP addresses because they have to pay for them because there are a limited amount.

With IPv6, this isn't an issue, and you don't have to worry about server farms having to use private IP space.  Google for example (if you look in your Google Mail/GMail e-mail headers) uses 10.0.0.0/8 IP space to number their servers, because not even google can be allocated enough space to number their servers with real IP space.  This causes them to have to worry about having NAT and other issues and a very complex network setup internally, instead of just using a global numbering scheme.

The problem isn't so much that every person with an Internet connection needs this many addresses.  The problem is that collectively we're running out of IPv4 addresses, so IPv6 was created (with some other new things added, not just more addresses).  When making IPv6, they had no reason to skimp on how many addresses to create, so they went large and went with 128 bits of address space.  If they just used 48 or 64 bits, it is more likely/more plausible that we'd run into the same exhaustion issue again in a finite amount of time.  IPv6 really makes this not a concern, at least not for a LONG, LONG time.

The fact that IPv6 allows for proper end-to-end connectivity as the Internet was originally intended is very nice, and it will completely eliminate the need for NAT (which creates headaches for anything more than a tiny network).