Talk:Classful network

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Computing / Networking (Rated C-class, Mid-importance)
WikiProject icon This article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
C-Class article C  This article has been rated as C-Class on the project's quality scale.
 Mid  This article has been rated as Mid-importance on the project's importance scale.
Taskforce icon
This article is supported by Networking task force (marked as High-importance).

switch or router?[edit]

the edit made on this date 21:58, 11 September 2006 seems awry.

they changed switches to routers and it seems to have stuck. note that they also changed the year from 1982 to 2006 etc etc.

Yeled (talk) 02:26, 24 January 2008 (UTC)


I'm pretty sure is the normal loopback address. Any ideas on resolving this? -- 15:56, 14 Mar 2004 (UTC)#

You are correct: is the normal loopback address. However, it's the lowest usable address in the classful loopback network which is reserved for local loopback, assuming the old rules of not using the all-zeros address (which requirement has been relaxed for CIDR). So both statements are valid. -- The Anome 16:02, 14 Mar 2004 (UTC)

Does anyone know what the Class E addresses (240.x.x.x-254.x.x.x) were reserved for? Experimental is what I seem to have found..anyone know what was meant by this?--Kcbnac 06:32, 13 October 2005 (UTC)

As a participant in IETF addressing work, I've never heard anyone explain the purpose of Class E, but there are so many implementations that would reject it as an invalid range, no one has ever really tried to make it work. Hcberkowitz 22:53, 25 June 2007 (UTC)

Not sure if I am right and guess this is the right place to bring it up, but I thought a class A address range ended at in (ddn) not as stated in the article?

the zero address range[edit]

An expansion of the purpose of the zero-address range by someone more in the know than I would be greatly appreciated. -- Jon Dowland 16:40, 16 January 2006 (UTC)

If you are speaking of subnet zero, that's an artifact of classful routing, in which you don't always have the subnet mask available but have to infer it from the first few bits. For example, without a subnet mask or CIDR length prefix, is a "class A network", or is it subnet zero of a 24-bit prefix? You can't tell by looking at the address alone.
With CIDR notation, the first is, but the second is There's no ambiguity for the all-zero subnet if you always know the mask. Hcberkowitz 22:51, 25 June 2007 (UTC)

I'm afraid of Americans[edit]

Don't be, it is the American Government you need to worry about.[Anonymous] —Preceding unsigned comment added by (talkcontribs)

I suggest you become a bit more familiar with the history of the Internet. In point of fact, the military proper had relatively little to do with the original development of the ARPANET and Internet, but provided funding to academic research organizations, including European and Asian organizations. I suggest you look at the rules at which IP addresses were originally allocated, and the present roles of the continental Regional Internet Registries, such as RIPE-NCC, APNIC, LACNIC, AfriNIC, and yes, ARIN. You seem to have an idea that there is some American organization that can give and take IP addresses away. It's not. Certain companies and universities, involved in research early on, did get large blocks. Some, such as Stanford University, returned their original large space, but at their own initiative. One of the reasons the allocation was wasteful is a point of this article: classful addressing is inefficient, but that's how initial allocations were made. Techniques such as Network Address Translation didn't exist for the ARPANET.
There are perfectly valid technical reasons to have blocks of registered addresses that are not accessible in the public Internet, such as for international telecommunications and financial organizations that are sufficiently large to need a block they can allocate to new members, without fear of duplication.
Classless Inter-Domain Routing (CIDR) enormously increased address allocation efficiency. In any event, whether or not organizations give up IPv4 address space, there are good reasons to go, systematically, to IPv6 addressing. The obsolescence of the IPv4 address space is not merely a matter of exhaustion, but of issues such as aggregation. Hcberkowitz 22:44, 25 June 2007 (UTC)

I have to chime in regards to the 2007 reference to ipv6. It's 2011 and we don't have anymore IPV6 now than we did in June 25 2007. I also don't understand how ip numbers that aren't accessable by the internet are useful to banks and other sectors. Bank of America has 5 or so million and I would entertain anyone to explain the need for that. Is someone going to say all the atms need an ip or 2? I also don't know how accurate the statement saying "nat wasn't around" is. While the term nat may not have been around the technology needed for deployment would of been no harder back then so im sure they did have it to some regard.

We might have large numbers of ipv6 being assigned to isps and organizations but none of it is being used in a production setting. I have it with hurricane electric and their ipv6 won't talk to level3's. What good is ipv6 when it won't talk to it's neighbor? All the hype and money invested in getting ipv6 out there apparently didn't also go to their peering relationships to make sure their ipv6 was of any use outside of their network. Woods01 (talk) 22:07, 20 March 2011 (UTC)

Suggesting that some orgs should give up addresses is not the same as saying that IPv6 is a bad idea. If the exhaustion of addresses is occuring more rapidly than IPv6 testing and deployment, then adjusting some of the wildly inefficient allocations seems like a natural way to bridge the gap. —Preceding unsigned comment added by (talk) 10:43, 17 March 2011 (UTC)


Say.. this article needs a couple of introductory sentences explaining what the hell it is about. As a general, non-tech, liberal arts reader, I have no idea what any of this article means. I understand that the more technical stuff is going to be incomprehensible to me, but perhaps just one or two sentences at the beginning to put it into context or something? Torgo 04:47, 31 March 2007 (UTC)

192.168 Class[edit]

What class is the 192.168 subnet? On the special ranges table it lists it as a class C with a /16 mask, while the "Class ranges" table says that a /16 netmask is a class B. Which is it? Same goes for 192.18, I guess... --Poromenos 19:43, 23 June 2007 (UTC)

With due regard that classes are obsolete, 192/8 (in current notation) or 192.x.x.x are class C. 192/8 is also called the traditional Class C space, or, among service providers, "the swamp". At one point, before CIDR, addresses in 192/8 took up half the global routing table, and prefixes of /24 or longer (i.e., mask,, etc.) took up half of 192/8. The latter part of the "swamp" was called the "toxic waste dump" in CIDR slang.
192.168/16 is designated as private address space, not routable on the Internet. is reserved for specialized testing.Hcberkowitz 22:48, 25 June 2007 (UTC)
While 192.168/16 appears to be a Class B subnet, it is actually a collection of 256 separate Class C subnets. The notations 192.168/16 is therefore somehow inaccurate, the correct approach would be to list the 256 subnets indivdually (192.168.0/24, 192.168.1/24, 192.168.2/24, ..., 192.168.254/24, 192.168.255/24).
This set of subnets was useful before CIDR appeared. With CIDR, the 192.168/16 range can indeed be used like a Class B subnet or whatever subnet sizes the user requires.
Before CIDR, some devices had hard-coded subnet class assignments. So they regarded 10/8 always as Class A, and it was not possible to split this subnet into smaller subnets. So 192.168/16 was created, for people who needed a number of small separate subnets. -- (talk) 15:25, 9 April 2008 (UTC)

End of Class A range wrong?[edit]

Shouldn't the end of the Class A range (under "Useful tables") be instead of -- (talk) 10:08, 21 January 2008 (UTC)

Oops.. 127.* is a special address range. -- (talk) 17:28, 21 January 2008 (UTC)

The correct Class A range is to, to is reserved per RFC 1700. —Preceding unsigned comment added by Phidjit (talkcontribs) 20:31, 12 March 2008 (UTC)

That's right so whatever retard put in that the range ends at needs to recheck their data. 127.x.x.x is reserved for loopback and does not have a class. this page is another instance of how wikipedia supports whatever people want to say based on whether they're wrong or not. —Preceding unsigned comment added by (talk) 20:23, 3 March 2010 (UTC)

The definition of class A addresses is that the leading bit is zero. That is the only criterion, and therefore 127/8 is a class A block. This has nothing to do with the actual use of the addresses. There are other addresses that are reserved as well, and they are still class A or whatever class addresses. No RFC has removed 127/8 from class A somehow. When a 127.* address is assigned to an interface it has always automatically received a /8 subnet mask. Kbrose (talk) 21:12, 3 March 2010 (UTC)

bits or bytes?[edit]

The third sentence in this article says, "Each class, coded by the first three bits of the address, defined a different size or type (unicast or multicast) of the network."

Perhaps that should be bytes (octet), not bits. —Preceding unsigned comment added by (talk) 01:51, 9 December 2008 (UTC)

No, it should say "coded by the first *few* bits of the address", see the first table which gives the leading bits for each class (0, 10, 110, 1110, 1111). Gvanrossum (talk) 20:37, 19 March 2009 (UTC)


Normally I'd just do a little gnoming on this, but I'm not sure I understand the subject sufficiently to be certain I'm not mistaken... "defined a different size or type (unicast or multicast) of the network."

Shouldn't "the" be removed from that sentence, or else the "defined a different size or type" be changed to "defined the size or type"? —Preceding unsigned comment added by Dr.queso (talkcontribs) 04:21, 30 June 2010 (UTC)

I agree the wording of the lead was somewhat confusing, and I have reworded it. Part of that was to change the tense from "defined" to "defines" because although classful networks are no longer used on the Internet, and indeed are probably extremely rare if not non-existent, there is nothing to stop someone setting up a classful network, so strictly past tense is clumsy and inaccurate. Johnuniq (talk) 07:20, 30 June 2010 (UTC)