Talk:Subnet

IPv4 Classes section needs tidying

There is some stray text in italics above the first table and the text (about loopback, etc.) below the table doesn't make sense.

IPv6

The article mentions /64 for residential customers, and yes for those with a single home network, that would be right. But some need more than one network in the home. Comcast, at least, will assign a /60 for residential customers to allow for that. I don't know that others do that, but as I understand it, that is the way it is supposed to be done. Should the article mention this? Gah4 (talk) 09:16, 2 March 2022 (UTC)

The whole idea

The article mostly misses the whole idea behind subnets and subnet addressing. That is, from outside the organization only the whole network address is needed. That greatly simplifies (reduces the size of) routing tables. Before subnet addressing, organizations would slowly collect more and more class C (/24) addresses, which would need to be added to routing tables everywhere. The other system used for subnetting, without subnet addressing, is Proxy ARP. In the case of proxy-ARP, hosts believe that they are on one large (often /16) Ethernet, as routers answer ARP requests fooling hosts into believing that the router is actually the destination. 100 or so hosts is about the largest you want a single Ethernet to be. Gah4 (talk) 09:35, 2 March 2022 (UTC)

Gah4, classes and classful networking were obsoleted in 1993. Subnetting has roots in that concept where the classful network was often subnetted before actual use. The exact same can be done with any delegated CIDR range. --Zac67 (talk) 10:37, 2 March 2022 (UTC)

Yes. Though I suspect some used proxy-ARP long past that. But again, the whole idea, either for proxy-ARP (that is, fake subnets) or subnet addressing, is to keep routing tables smaller. From the outside, a large organization has one network address, which looks like the classful network. (Most often a class B network.) Before about 1988-1990, many larger organizations had a large number of class C addresses. The fix for that was, in many cases, a class B address (from the outside) and subnetting from the inside. CIDR came just a little later, when people started to worry about running out of class B addresses. Also, with CIDR, splitting up the class A addresses, which were too big for most organizations. Gah4 (talk) 21:17, 2 March 2022 (UTC)

OK, until about 1992, NSFNET was the main backbone of the US network. I suspect it is mostly forgotten by now, but NSFNET was government sponsored to connect US government and academic institutions (especially ones with government funding). Commercial users were allowed, but only for connections to/from government and academic users. You were not allowed to email or ftp from one .COM to another! There were no large ISPs like today, so the routing tables had to be able to connect everyone to everyone else. (Much of this is in NSFNET.) As large commercial ISPs formed, about after 1992, they allowed commercial traffic on their nets, and kept it off NSFNET and other government funded nets. But also, this means that much routing is inside a large ISP, which can then allocate IP addresses, using subnet addressing, to minimize routing table size. And also, that is when CIDR appeared, as that is when it was needed. So now, a big ISP might start with a /8 net, and divide it up among large and small users. Also, addresses are divided up between countries, again to keep routing tables small(er). So, it is the small(er) number of large ISPs, and CIDR, that keep things working. Gah4 (talk) 22:11, 2 March 2022 (UTC)

Other than IP

I used to know about Appletalk phase 2, but haven't thought about it for a while. There is also DECnet. It would seem that subnets are a good idea, and that other network systems would also use them. Gah4 (talk) 02:55, 7 February 2023 (UTC)

Requested move 28 June 2023

The following is a closed discussion of a requested move. Please do not modify it. Subsequent comments should be made in a new section on the talk page. Editors desiring to contest the closing decision should consider a move review after discussing it on the closer's talk page. No further edits should be made to this discussion.

The result of the move request was: moved. Subnet has been agreed upon here as the common name (closed by non-admin page mover) ASUKITE 15:09, 13 July 2023 (UTC)

---

Subnetwork → Subnet – Per WP:COMMONNAME. PhotographyEdits (talk) 11:17, 28 June 2023 (UTC) — Relisting. Zippybonzo | Talk (he|him) 15:21, 5 July 2023 (UTC)

• Support. --Zac67 (talk) 17:29, 28 June 2023 (UTC)
• Oppose Do we have a WP:RS that subnet is the WP:COMMONNAME? I believe it is commonly used as an adjective, as in subnet mask. And maybe also in casual usage. Gah4 (talk) 21:56, 28 June 2023 (UTC)

  @Gah4 Yes, here is the evidence. PhotographyEdits (talk) 07:49, 30 June 2023 (UTC)

• Oppose The full title is much more intuitive for general audiences, rather than jargon used by specialists. Acronym and abbreviation mania should not be practiced in encyclopedias. kbrose (talk) 16:15, 29 June 2023 (UTC)

  That's simply not what the policy of Wikipedia is, take a look at MOS:ACROTITLE. PhotographyEdits (talk) 07:49, 30 June 2023 (UTC)

• Comment I believe it is subnet as an adjective, such as subnet mask, and as noted likely in casual use or jargon. Titles are supposed to be nouns, not adjectives. Gah4 (talk) 01:58, 30 June 2023 (UTC)
• Oppose The correct word is "subnetwork" and that is the encyclopedic term that should be used. Of course people use subnet as an abbreviation but no evidence has been presented that it fits COMMONNAME. In a similar spirit, there are a zillion books on networks and subnetworks and subnetting, but not so many on subnets. Johnuniq (talk) 02:06, 30 June 2023 (UTC)

  @Johnuniq Look at my comment above for the evidence. PhotographyEdits (talk) 07:50, 30 June 2023 (UTC)

  An n-gram is not evidence of how words are used in titles. Once the topic of subnetworks has been introduced in a book, it is common that the term be abbreviated to subnet. The fact that an abbreviation is frequently used is not a reason to use that abbreviation for the title. Should Celsius be renamed to its abbreviation? Johnuniq (talk) 08:04, 30 June 2023 (UTC)

  Celsius is commonly used in full in daily usage. Subnetwork isn't.

  The n-gram does indeed not provide evidence for how it is used in titles, but the titles of Wikipedia should reflect the usage in the cases that n-gram measures. So it is good evidence. PhotographyEdits (talk) 10:52, 30 June 2023 (UTC)

• Support - even in layman contexts, such as the instructions given when you by a wifi router or the networking settings visible on your own computer, the term used is "subnet". It is not jargon, and in fact the current title is far more "jargon-y" than the alternative simply because a reader would nearly never encounter the term "subnetwork" and might think this article is the wrong destination. WP:COMMONNAME has to prevail here, as the arguments opposing are too spurious to override it. -- Netoholic @ 11:37, 30 June 2023 (UTC)
• Support per Netoholic; I've only ever heard of "subnets", and "subnetwork" is hardly the correct word here -- rather, it is a sort of hypercorrection. Even the RFC 950 refers to it as "subnet", along with the vast body of literature. No such user (talk) 11:37, 3 July 2023 (UTC)

Support per Netoholic and No such user. We title our articles per WP:COMMONNAME, not per WP:ENCYCLOPEDICNAME. —В²C ☎ 04:57, 5 July 2023 (UTC)

Support. I don't believe the evidence supports the suggestion that attested use is as an adjective. Collins, for example, clearly indicates that it is a noun. The Free Dictionary (AHD) also has it as subnet, though Oxford and Merriam-Webster have it as subnetwork. Most other dictionaries (MacMillan, Cambridge, etc) do not seem to have either. So, even split there. Etymologically, per No such user there does not seem to be convincing evidence that the shorter form is derived from the longer, unlike internetwork. There is, of course, RFC-932 which precedes that, but RFC-917 precedes that in turn. It seems equally likely in my opinion that the two forms were derived in parallel, however much significance we should ascribe it. I do note that an intitle: search dies yield more books with "subnetwork" however it does appear that not all of those are about computing. Overall, I find it highly likely that the shorter form is much more common in its own right rather than as an abbreviation, however, I am open to revising my opinion should anyone be willing to do a more detailed review. Alpha3031 (t • c) 15:36, 5 July 2023 (UTC)

• Support. Perhaps in some people's intuition it's inescapably obvious and logical that a word "subnet" which has to do with networks is necessarily just an abbreviation of "subnetwork", but it's not clear that it's the correct/unique analysis, as opposed to < sub- + net (as found in the OED; it acknowledges a relationship with "subnetwork" but does not call it an abbreviation or even a shortening). And even if it is an abbreviation, that doesn't preclude it from being the best title. The word "subnet" (a noun, not an adjective) is common and recognizable, as other editors have discussed above, and commonly used without ever introducing it alongside "subnetwork". Adumbrativus (talk) 05:04, 6 July 2023 (UTC)
• Support - I personally prefer the current title because it is less jargony but WP:COMMONNAME says otherwise. ~Kvng (talk) 13:02, 8 July 2023 (UTC)
• Support per nom. As to two opposing comments: Subnet is clearly used as a noun, adjective, and gerund based on even a brief glance at Google Books results. And I am not any sort of specialist, but I have only seen "subnet" when interacting with WiFi settings, not "subnetwork". (That's just based off my experience accessing the Internetwork.) {{replyto|SilverLocust}} (talk) 10:00, 13 July 2023 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.