Talk:Subnetwork/Archive1

From Wikipedia, the free encyclopedia

This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Contents 1 Should protocol-specific details be in here? 2 Error in decimal->binary conversion? 3 Some history 4 Wrong example? 5 Adresses starting with 127

Should protocol-specific details be in here?

This article currently contains strange paragraphs like

"A network mask, also known as a subnet mask, netmask or address mask, is a 32-bit bitmask used to tell how much of an IP address identifies the subnetwork the host is on and how much identifies the host."

Which is, of course, completely false in the general sense. Has the author not heard of IPv6 for example?

... which raises this question: Should this article really have anything to do with specific protocols, or should it just be an explanation of the term subnetwork with pointers to specific articles about IPv4 subnetting etc? We already have the IPv4 subnetting reference so that seems logical to me.

I would understand links to specific articles from here (such as one for IPv4 subnetworks and another for IPv6 subnetworks), but this seems a bit illogical. How about lifting protocol-specific material out to the revelant articles and pointing to them?

When e.g. IPv6 becomes more common, we'd otherwise have to include that one here as well to be consistent. It's of course another option, but again, we already have a specific IPv4 article for this -- why not use it better? Just a thought. :-) Jugalator 18:39, Aug 21, 2004 (UTC)

My primary reason for clustering them together was too many sub-stub-articles were created as a result, all of which were being pointed to by different articles, usually intending the same (IP specific) idea (Ex: Subnet, subnet address, subnetting, subnetwork, Classful network). I just took the vast (and often quite vague) array of IP-Specific Subnetting articles, and crammed them into one larger, easier to understand page.

The protocol-specific information could be difficult to remove, since different protocols handle subnetting differently (if at all), and reducing it to an explanation of what subnetting means, might again reduce it to a stubby dicdef.

Still remaining is a slew of other articles on the topic Internet_Protocol, IP address, IPv4 subnetting reference as well as IPv4, IPv4 address exhaustion and IP address allocation to boot.

We could simply rename the article to something like IPv4 Subnetting (IMHO adding the word "reference" may be a bit too verbose), make a redirect, and hope that someone working on one of the other (poor, abandoned) Network protocol pages (like AppleTalk, IPX or *gasp* Xerox_XNS) makes an appropriate disambiguation page if nessecary. Gamera2 06:28, 23 Aug 2004 (UTC)

Follow up

It seems as though IPv4 and maybe IPv6 (The same thing, minus a few extra features and changes for the sake of easy conversion), are the only two protocols that use "subnetting" in the flexable sense. All the other protocols (or at least the ones anyone knows anything about, can't speak for BanyanVINES myself), appletalk or IPX for example, don't seem to support any kind of masking past the pre-set network half of the address (for obvious reasons). If someone wants to put in a paragraph (or change the top one accordingly) to describe network/host halfs of logical addresses, then they're more than welcome to do that. Maybe add tidbits about how it's handled in IPv6, etc. - Gamera2 18:23, 13 Oct 2004 (UTC)

I started to try and clean this article up before I saw all this. I don't think we should have separate article for IPv4 and IPv6 subnetting - after CIDR they are now close enough in mechanism that there's no point. It would result in several tiny pages - something the Internet Protocol area has enough problems with already. IPv4 subnetting has a lot of history IPv6 subnetting doesn't have, but other than that they are now the same. If people want to split this page up, and have a page called "IP subnetting", that would be fine - but the remaining material on the subnet page would only be a few sentences.

As for the Internet Protocol mess, I've been working on cleaning individual pages up, but we could probably use some rational design of how many pages we're going to have, and what's in each. Do we want to start a WikiProject page to coordinate this, rather than having comments scattered here and there on various Talk: pages? Noel 02:57, 27 Oct 2004 (UTC)

Actually, I was looking for technical-specific information. Perhaps that could be added in a section, albeit probably very large? To help readers understand better, perhaps an example, one being an overview another technical, of information traveling through the internet as a user browses webpages through their home network (ie: through a router). Great article nonetheless! 68.3.8.223 15:51, 25 March 2006 (UTC)

Error in decimal->binary conversion?

In

"Network address 204.4.32.0 Decimal = 11001100.00000100.00100000.00000100 Binary"

shouldn't the last eight digits be all zero?

Snip 14:11, 9 September 2005 (UTC)

Yes. Noel (talk) 17:52, 19 September 2005 (UTC)

Some history

Putting this here as data for anyone who cares:

The first mention of subnets, in the sense of subdivisions of a classful network, that I know of in Internet documentation occurs in IEN-82, "LCS Net Address Format", from February, 1979. (MIT was using subnets some years before anyone else.) This subnetting scheme, as eventually adopted by the Internet, was more fully described by Jeff Mogul in RFC 917, "Internet subnets", in October 1984.

The notion of subnet masks has to be credited to Dave Moon, though, I think. Although the early LNI hardware supported masks, we didn't really think of using them in the protocol; it was Dave Moon, at an early meeting on an otherwise forgotten piece of technogical detritus named 'MUPPETS' (the name is a play on PARC Universal Packet) - an attempt to deal with the multiplicity of protocol suites inside MIT at the time - who made the mask suggestion at the protocol level, and it was carried forward to Mogul's paper. Noel (talk) 17:52, 19 September 2005 (UTC)

Wrong example?

In the example of the chapter "subnetworking concept" it says:

"Determining the number of hosts and subnets on a particular network is quite easy, if you know the subnet mask. Say you have the network address 154.4.32.0 with a subnet mask of 255.255.224.0. This network address can also be written as 154.4.32.0/19

Network address 154.4.32.0 (decimal) => 10011010.00000100.00100000.00000000 (binary) Subnet mask 255.255.224.0 (decimal) => 11111111.11111111.11100000.00000000 (binary) The subnet mask has 19 bits for the network portion of the address, and 13 bits for the host part.

2¹³ = 8192 possible subnets available according to RFC 1812, otherwise using the old RFC 950 standard the number of usable subnets is 6. This is due to RFC 950 (section 2.1, page 5) not supporting subnets with either all 1s or all 0s."

Shouldn't this be 2¹⁹? - Otherwise, an explanation of why a mask of 19 bits only gives us 2¹³ possible subnets would be nice. (Admitted - I did not read the RFC's, which is also why I didn't just start editing the article on my own, but I think, this explanation goes against a basic sense of math...? —Preceding unsigned comment added by 62.107.132.247 (talk • contribs) 18:46, 11 July 2006

The length of the netmask is 19 (out of 32), so the length of the network part if 19 and the host's is 32 - 19 = 13. Right?

Why 2³ = 8? I mean, why 3? Should we assume this is a class B (/16) network, and we're trying to fit that many /19 in it? So 19 − 16 = 3?—The preceding unsigned comment was added by 82.238.35.175 (talk • contribs) 20:47, October 22, 2006.

Adresses starting with 127

The current article says that The 127.0.0.1 Network ID is left out because it is designated for loopback and cannot be assigned to a network. Are there really not more reasons to be mentioned here? It looks very inefficient to take out 2²⁴ IP-adresses, and then use only one of them. Bob.v.R 09:03, 15 May 2006 (UTC)

127.* is loopback; I can't see any other reason if all the range already has a known and common meaning. Actually, it's not entirely true that only one of them is used. You can use whatever IP in the range to refer to loopback (and I guess using different IPs has some use). There're other inefficiences in the IP, so why not this one :P --Outlyer 17:48, 15 May 2006 (UTC)