Talk:IPv6/Archives/2009

The Pirate Bay announces IPv6 support
Perhaps not yet but in the very near future: http://thepiratebay.org/blog /Tense (talk) 14:18, 24 January 2009 (UTC)

IPv6
I have encounter a question regarding this. Sorry but this topic is new to me. Though i've known IPv6 for quite some time, technically, i don't know much about it.

Some of my questions:

Are all IPv6 starts with [2001::]?

Ragarding loopback, [ ::1] is this the only loopback address for IPv6?

and lastly, when the IPv6 address is assigned to an interface, does this IP is dedicated only for that interface and can not be used by others?

Thanks you very much for anyone who can answer my question.

-teknika —Preceding unsigned comment added by Teknika boy (talk • contribs) 12:34, 14 March 2009 (UTC)
 * Talk pages are for discussing the article, not for asking questions, as per WP:NOTFORUM. As discussed in the article, not all IPv6 addresses start with 2001::.  As discussed in the article, only ::1 is the loopback.  I'm not sure what your last question is really asking, but I suspect the answer would be too long to address here.  Wrs1864 (talk)

Contradiction in Simplified processing by routers
IPv6 routers do not perform fragmentation. IPv6 hosts are required to either perform PMTU discovery, perform end-to-end fragmentation, or to send packets smaller than the IPv6 minimum maximum transmission unit size of 1280 bytes.

I may be wrong, but this sounds to me like a contradiction. If the minimum MTU is 1280 bytes we must not send smaller packets. Since I'm only starting reading about Ipv6 I'm cannot tell yet what's correct.

84.57.146.147 (talk) 10:39, 29 March 2009 (UTC)


 * This not a help forum, but you should examine the assertion 'we must not send smaller packets'. How else would you send 10 bytes of data? The minimum MTU specifies the smallest MTU a link must support. Has nothing to do with the amount of data to send (IPv4 or IPv6), but the limit until which data is transmitted without fragmentation. Kbrose (talk) 14:00, 29 March 2009 (UTC)

Vague section on "users disabling IPV6" is unencyclopedic
This section needs to be be removed immediately as it is vague, and reads like a "tips" self-help guide, not an encyclopedia entry about IPV6 itself. The important points in it should be moved to and integrated into the sections on deployment issues, but only if they can be made far more concrete and specific with citations to the detailed problem types that have been documented.83.105.29.228 (talk) 20:51, 14 April 2009 (UTC)
 * I tried to address your concern by rephrasing the whole section without giving the impression to offer advice of 'disabling'. Still would be nice to have some explicit references though. Kbrose (talk) 18:25, 15 April 2009 (UTC)

Clarification - the host identifiers are decoupled from ...
The article (section 2.1) says: "With IPv6 [..] the host identifiers (the least-significant 64 bits of an address) are decoupled from the subnet identifiers and the network provider's routing prefix." Does it mean that the host identifiers are unique? That there is maximally one IPv6 address with a particular host id? If yes, then it should be said. If no, can it be clarified what is meant by "decoupled"? --Jirka6 (talk) 22:47, 14 April 2009 (UTC)
 * I hope I clarified the issue. Kbrose (talk) 18:25, 15 April 2009 (UTC)

Accuracy
The article needs to be more accurate in the number of possible IP addresses because computer scientists are interested in this information. 207.160.210.187 (talk) 00:36, 18 April 2009 (UTC)

First statement
Gentlemen,

when some internet users seek articles herein, they want (and need) to learn the easiest way.

We shall not consider even the hardest article to be read only by those with enough knowledge.

Besides confusing layers with protocols untill I corrected it, the first statement serves both ipv4 and ipv6 concepts but it gives the idea it is only about ipv6:

"Internet Protocol version 6 (IPv6) is a network layer IP standard used by electronic devices to exchange data across a packet-switched internetwork."

Perhaps we should think about it.

as of may 2009
first sentence seems a lot better. Although it's wordy, you have to convey the facts, and there's enough alternate words. Naive translation: Internet Protocol version 6 (IPv6) is the next-generation Internet ... protocol for ....the Internet.

The second sentence introduces IPv4, the 'old' version, giving context.

I think the 'too technical' label is really old and it's been fixed. I'm taking it out. OsamaBinLogin (talk) 18:53, 16 May 2009 (UTC)

Motivation for IPv6
In my opinion, the sentence "Incidentally, the IPng architects could not use version number 5 as a successor to IPv4, because it had been assigned to an experimental flow-oriented streaming protocol (Internet Stream Protocol), similar to IPv4, intended to support video and audio." is of no interest in here. It is just confusing. —Preceding unsigned comment added by 88.197.223.54 (talk) 12:24, 26 May 2009 (UTC)
 * I don't see why it would be confusing, it is a pretty clear unambiguous statement. It is also of interest and relevant in context, as the question 'why version 6 instead of 5?' comes up all the time. Kbrose (talk) 15:33, 14 June 2009 (UTC)

Stateless Addr Autoconf
From the article: | Stateless autoconfiguration is not used by routers. The sentence could be read as "routers do not provide stateless autoconfiguration" by non native speakers (like me).

In my opinion, this should be rephrased to Stateless autoconfiguration is not used by routers to obtain their own address to make it clearer. I'm not sure if my suggestion is correct though as I haven't read RFC 2894. --Gregor Alessi (talk) 13:40, 14 June 2009 (UTC)
 * Thank you for your observation and input. The statement indeed is not only confusing, as it was written, it is simply false, at least excessively oversimplified. I think its origin may have been the old IPv4 philosophy of assigning fixed addresses to routers. The IPv6 picture is much more complicated, and routers can indeed use a form of address autoconfiguration provided by RFC 2894. In addition, link-local addresses may be assigned in routers by the identical internal mechanism used by normal IP hosts. Kbrose (talk) 15:27, 14 June 2009 (UTC)

Interesting historic parallel
Here is an interesting reference from our Rail gauge in North America article. It describes the change in railway gauge throughout the southern United States from 5 foot to 4 foot 9 inches (essentially standard gauge), which took place over the course of just two days, starting May 1, 1886.[ http://southern.railfan.net/ties/1966/66-8/gauge.html]--agr (talk) 19:44, 26 August 2009 (UTC)


 * It's not a fair comparison. The train system was a centralised, organised system controlled by a small number of organisations.  This is very much unlike the Internet, which is a chaotic, dis-organised, decentralised system. --Jec (talk) 09:08, 30 August 2009 (UTC)

The size of the IPv6 address space
In the article, the size of the IPv6 address space is attributed to be "much larger" than the IPv4 address space, without any hint to the reader that this is a huge understatement. The reader is misinformed furtheron when shown the image of the growth of the address space. The image represents the address space of IPv4 and IPv6 as two equally sized rectangles and lets the viewer believe that the IPv4 address space is represented by a single pixel in the square representing the IPv6 address space. It even has a comment on the side of the image stating that "image is to scale, except the black area is enlarged for ease of viewing". The "black area" in the IPv6 square is enlarged about 1036 times to be big enough to be visible!

Some quick calculations learn that you could operate more than 13200 complete IPv4 networks (having 4,294,967,296 addresses each) on each square millimeter of the Sun's surface, which is 11,990 times the size of earth's surface.

I think this kind of "BIG" needs to be expressed more strongly in the article and the image representing growth should be removed since it is misleading.

--Dandorid (talk) 19:55, 13 September 2009 (UTC)
 * Despite the drastic increase in address space from a purely numeric point of view, this should not be overstated or compared directly to IPv4's space. The actual space available, while notable, is not so important other than that it is deemed sufficient for the foreseeable future. The address space size has to be considered in conjunction with the addressing architecture used in the new system. Subnet's are always 64-bit subspaces, which are in size the square of the entire IPv4 Internet, and even residential networks should be allocated a block of 216 (~65,000) subnets according to the plan. So, even a small home network of, say, half a dozen devices, will use up an address space 'vastly' larger than today's Internet. Thus the allocatable address space at the registry level (the number of subnets that can be delegated by the registries) is much smaller than what the numbers may indicate. It is simply silly to compare IPv6 next to IPv4 space and calculations as above are even sillier. Kbrose (talk) 06:22, 18 September 2009 (UTC)

Last edit summary
Sorry about that, that was supposed to read "Fixed link to only hit of 'IPv6' on Limelight website.", but I accidentally hit enter while aiming for '. -pinkgothic (talk) 10:32, 28 September 2009 (UTC)

IPv4 is to IPv6 as octet is to...quad?
It was useful in real life conversation to call portions of the IPv4 address "octets". Since IPv6 address portions are broken up into 4 hex characters, can we use the term "quads"? Discuss - milliamp
 * They are called nibbles. —— Da n do r iD (talk) 18:49, 25 October 2009 (UTC)
 * Oops, a nibble is four bits not four bytes. So discussion still open. I personally don't feel the need to invent a name for these groups of four hex digits; 'group' will do just fine. —— Da n do r iD (talk) 11:53, 28 October 2009 (UTC)

A rather serious problem with this article...
IPv6 is a major development and although in its infancy of adoption, appears likely to become high profile and widely used over time (from IP v4 exhaustion if nothing else).

The lead article on IPv6 should ideally be structured like the general article to any other major topic - sections covering each aspect in overview with more detail in separate articles.

Instead it's mostly technical detail and specification, which would do well moved to an article Design and specification of IPv6, but for the ordinary person who has heard of IPv6 and want to know in everyday terms about it, there's very little information. Even the article on IPv6 deployment is more a recital of what major organizations are deploying it in each area.

A simple example - how do routers and devices negotiate their communication when both IPv4 and IPv6 are possible? What changes to networking norms, subnets, and the like, are expected as a result of IPv6? If a person or organization considers moving from IPv4 to IPv6 what impact does that have (apart from replacing some legacy hardware)? What sort of considerations might arise?

Some major reworking of this article might provide a really first class introduction and reference page on IPv6. Any chance of reworking the article to that end? FT2 (Talk 04:01, 18 September 2009 (UTC)


 * I fear that your simple example is not so simple to explain in layman's terms, at least not when avoiding techicalities. Due to all sorts of technical stuff you end up having multiple ways of how this could work. That would undo your strive to make this a non-technical article. I agree that the general IPv6 article should contain a general text describing the outline and idea of IPv6. It can then link to a series of technical IPv6 sub-articles. Is this the way forward? Da n do r iD (talk) 11:59, 20 September 2009 (UTC)


 * Tread lightly. I've seen this article used by a lot of people, mostly network admins who've never worked with IPv6 and who suddently find themselves needing to learn in a rush.  So be very careful to make sure the article remains useful even when in a state of transition, and be careful not to introduce any incorrect information.  In particular, the subpages should be created first, and only when they have been vetted by others can the text from the main IPv6 page go away. --Jec (talk) 18:55, 22 September 2009 (UTC)


 * I have started a draft for a sub-page on my user page: user: dandorid/IPv6 Addresses. I will notify when it is ready for reviewing. Da n do r iD (talk) 18:46, 29 September 2009 (UTC)


 * My first draft for the new article "IPv6 Addresses" is ready for review; please do so. It is a collection of sections from the main page, in a more or less logical order, with minor modifications. Anything there is to say about IPv6 addresses itself is (or should be) in this article. After moving this to the main space most of the main article's sections on addresses (which are now scattered all over in the main article) can be removed, if they are replaced by an general piece of text about addresses which references the new IPv6 addresses page. When all sections about addresses are gone, a more comprehensive text remains, dealing with IPv6 itself. This text can then be restructured.  Da n do r iD (talk) 21:20, 4 October 2009 (UTC)
 * Wow. That's all I can say. This looks (to me as someone knowing nothing whatsoever about IPv6 but a reasonable amount about IPv4) to be a thorough summary of how addressing works in IPv6, and addressing changes compared to IPv4. It doesn't yet "explain" IPv6 but as you note above it's a big topic and perhaps creating sub-articles for its main themes first, will simplify that task. I cannot judge its technical merits but it looks ready for mainspace and for linking into the present article.


 * Also agreeing with Jec that usefulness must not be lost, which in practice may mean sub-articles like this are needed, then the relevant sections in the main article can be linked to these for more detail. That way, the main article can cover a more balanced comprehensive ground while retaining all the useful technical information on IPv6 that's included at present. FT2 (Talk 11:47, 7 October 2009 (UTC)


 * Next step, I guess, would be rewriting part of the main text into a summary text linking to the new page. I will try to do so, again on my user page. If you have a better idea please let me know.  Da n do r iD (talk) 09:57, 9 October 2009 (UTC)

Good work for IPv6 Addresses (and I'm proud to see that you've decided to keep most of my prose). Here's a first proof-reading:

Other than that, this page looks ready to be made into a real Wikipedia page.--Jec (talk) 03:22, 12 October 2009 (UTC)
 * the introduction needs more work;
 * Networks: as a further shorthand: no, that's not standard notation. You should clearly say that it is merely a non-standard notation used by some Linux tools.
 * IPv4 mapped addresses: the notation is not an exception -- using decimal for the last 4 octets is legal with all IPv6 prefixes. This should be moved to the previous section, with a forward-reference (something like "usually used with the IPv4-mapped notation, see below").
 * IPv4 mapped addresses: do all of these details belong here? This is a page about addresses, not about transition mechanisms or socket APIs
 * Literal IPv6 addresses in network resource identifiers: the third sentence (the one with however in the middle) doesn't sound right to me.
 * This is not only useful but mandated when using the short form: I was under the impression that the brackets are always compulsory. Please check.
 * Unicast addresses: please remove the discussion of scope from here, and move it further down.
 * Link-local addresses and scope indices: please remove the last paragraph (relatively few applications), it is no longer correct.
 * IPv6 addresses in the Domain Name System: please remove the last paragraph (about A6 and DNAME), it is obsolete.


 * First of all, thanks for the positive feedback. Second, I have kept most of your prose, as I did for all other pieces of text. Most of it was copying text from the main article and finetune them into the shape it is now.
 * As for your comments:
 * Introduction: true, needs a more comprehensive piece of text.
 * Networks: This sentence should be rewritten (I'll see what I can do), but I disagree that the notation is non-standard. RFC 4291 clearly states this way of representing (full) address and prefix (which then reveals the network the host is in). See paragraph 2.3. Maybe I am missing your point...
 * IPv4 mapped addresses/notation: You may be right, but I guess it is not very helpfull making people believe it is customary to use this notation for addresses other than ::ffff:d.d.d.d addresses. It may be a legal notation but does not make much sense. We can leave it open: just say that this is an alternative notation that is used for IPv4 mapped addresses.
 * IPv4 mapped addresses/technical stuff: I agree, does not belong here. But to delete it entirely is a waste. Should be kept on the main page as a separate feature of IPv6.
 * Literal IPv6 addresses in URLs: This piece of text is a mess... The references to RFC's are wrong (3986 supersedes 2732) and brackets are mandatory (RFC 3986 sec 3.2.2). I will rewrite it.
 * Address Scopes: this should be a separate section I guess.
 * Link-local: will delete text.
 * DNS: Should be mentioned on the Wiki about DNS. More work there... —Preceding unsigned comment added by Dandorid (talk • contribs) 18:35, 13 October 2009 (UTC)


 * Networks: ah, I didn't realise that. Looks like you're right.
 * IPv4 mapped addresses/notation: then just mention that it's customarily only used with v4-mapped in v4-compatible addresses.
 * DNS: here's a threat: unless you dump the A6 information, I'll add a section on addressing in SIP ;-) --Jec (talk) 01:26, 15 October 2009 (UTC)


 * Second draft of IPv6 Addresses
 * Well... A lot has happened to IPv6 Addresses. I have moved a lot of text around and added to most of the sections. I am quite happy with the new introduction I have written.
 * As almost all text has been touched (more or less) a new and thorough review is needed. I also could use a hint as how to move the text to the main space. —— Da n do r iD (talk) 23:59, 24 October 2009 (UTC)

I've just had a quick glance, it looks correct except for the fact that the last paragraph (about Microsoft's hack) doesn't belong here. I suggest you make your page into a real wikipedia page so we can edit it there.--Jec (talk) 01:37, 2 November 2009 (UTC)
 * Moved IPv6 Addresses to main space as you suggested. Discussion of the text can now be done on its Talk page. Maybe we can close this discussion, as rewriting of this article is underway as requested...  —— Da n do r iD (talk) 12:19, 2 November 2009 (UTC)
 * Ack, moving to Talk:IPv6_Addresses.--Jec (talk) 20:41, 2 November 2009 (UTC)