Talk:Classless Inter-Domain Routing
From Wikipedia, the free encyclopedia
Contents |
[edit] Prefix/suffix
I'm not quite sure here, but shouldn't "prefix" be replaced with "suffix" instead? --Maik
I think the usual terminology is prefix, since you route packets destined for 193.137.7.30 (for example) through the shortest path to the 193.137.7/24 prefix (which itself is aggregated and reachable through the 193.136/15 prefix (which is composed of the 193.136/16 prefix and the 193.137/16 prefix)).
The term prefix makes sense when you consider how the netmask is expressed in binary Robertbrockway 04:40, Apr 18 2005 (UTC)
I added a new diagram which should help clarify this.
Baccala@freesoft.org 06:31, 9 November 2005 (UTC)
[edit] Confusion
How in the heck did I interpret 4 minutes as 3 months??!? - Lucky13pjn 19:48, Mar 14, 2005 (UTC)
[edit] Example
The example about the /22 mask lists a Class-C address (192.168.0.0) but the default mask for Class-C is 255.255.255.0 or /24. I thought that you could only add ones bits to the default mask, not take them away. Wouldn’t a /22 mask have to be associated with a Class-A or B address?
This is the example given:
“192.168.0.0 /22 could be written 192.168.0.0 255.255.252.0”
Should it have been something like:
“172.168.0.0 /22 could be written 172.168.0.0 255.255.252.0”
--addnet 18:51, 23 July 2005 (UTC)
The restriction on taking bits away from the default mask only applied in the classful networking scheme. Classless routing is just what its name implies — you really don't have any restrictions anymore on what can be done to a "class A" or a "class C" address. With the exception of class D/E, the entire address space is now administered uniformly throughout. So there are no more "classes". Really.
Baccala@freesoft.org 06:29, 9 November 2005 (UTC)
Todd Lammle in his book CCNA: Cisco Certified Network Associate Study Guide, Fifth Edition ( San Francisco, London: Sybex, 2005) clearly states on page 109 that you cannot change the default mask. The address space is still classed because of class A, B and C network/broadcasting addresses. So a Class C network cannot have a submask shorter than /24 because the X.X.X.0 is the network address for it.
81.20.147.194 17:08, 25 February 2007 (UTC)
[edit] Origins of CIDR
The first person to suggest using a net-mask to cover a group of classful networks (as opposed to subnets of a single classful network) was Carl-Hubert Rokitansky (affectionately known as "Roki" to all :-), in his "Cluster Networking" proposal (which long predated CIDR). (See this message and this reply to the internet-history mailing list.)
I had thought for a long time that there were no original documents on line for this proposal; web searches revealed nothing. However, I have just discovered that the online copy (large [[Portable Document Format|pdf] file) of the Proceedings of the '1st' IETF (there was actually at least one earlier one) includes a copy of Roki's handwritten slides on the matter, on pp. 45-61 (pdf page numbers). Just noting this here for informational purposes! Noel (talk) 16:11, 16 September 2005 (UTC)
- I always thought the story of the Jan 16, 1986 meeting as the "first" one was the only version... but since it's Noel talking, I suppose not... Noel, can you correct the story on IETF? --Alvestrand 17:06, 16 June 2006 (UTC)
[edit] Order, Order
I like the way you use a notation ie : " ... sixteen contiguous /24 networks ... " which is meaningless until you read later what the / means. Maybe you could assume the reader is new to the ideas of Classless Inter-Domain Routing, which is why he/she came to the page.
Similarly " ... The class, and hence the length of the subnet mask ...", before explaining what a subnet mask is. Very odd. I assume you are writing this for the people who already know it all, and just need a reminder?
Point taken. I think it's better now.
Baccala@freesoft.org 06:30, 9 November 2005 (UTC)
[edit] External links
There are plenty of online network calculators, and the two ones in the external links are quite limited in functionalities. How about adding this one. --Olivier Debre 07:45, 14 April 2006 (UTC)
And how about adding this: http://vlsm-calc.net/. 80.250.189.67
[edit] Assignment_of_CIDR_blocks
On the number available on the 208.130.28.0/22 block, it states it is just above 1000. I was about to precisise it to 1022 (ie, 2^10 - 2), to addresses lost to multicast and to loopback, but I wanted to place the suggestion here for a while, just to check if I am wrong. Comments welcome. -- Heptor talk 19:55, 17 May 2006 (UTC)
[edit] A little plain as it lacks discussion of drawbacks
I do believe there has to be something we sacrificed after CIDR from the original Internet. After all, there is no decision in life that doesn't have opportunity cost. I do believe Internet is less resilent to failure due to the use of CIRD. See, before CIDR a fault would take down a couple of networks, now it can take out countries service. Such info would improve(make the article stand out) the article as its rarely discussed in other written work
- Could you elaborate a bit on this? It seems to me that in the "old days" of the Internet, when you had an entire Class C allocated to a user with ten hosts, that it was less resilient than if you take that traditional Class C space and chop it up into more useful and flexible chunks, which CIDR allows. CIDR, in and of itself, won't take down a country's service (this is what I think you need to expand upon). A bad implementation of BGP, on the other hand, could do this. Akamantauskas 21:10, 31 January 2007 (UTC)
[edit] VLSM?
CIDR and VLSM are basically diferent topics. whenever I ask for VLSM, it automatically directed to CIDR. Whats the matter ?????????????????
Sushil Kumar, India
- (moved new discussion to bottom of page) Someone must have redirected VLSM to point here. I don't know what VLSM is intended to be an acronym for, so I don't know if it's right or not. If you want to, just click the "Redirected from" link and start editing. --Alvestrand 16:18, 16 June 2006 (UTC)
-
- Variable Length Subnet Mask(ing) Luteijn 23:08, 11 July 2006 (UTC)
[edit] Regarding Prefix
I am cofused whether the prefix would be no. between 0 to 32 or 0 to 31!!
- The numbering is from 0 to 31, as IP addresses are 32 bits in length, and there are 32 numbers between 0 and 31 (count 'em). If you did 0 to 32, you would end up with 33 total numbers. Akamantauskas 21:06, 31 January 2007 (UTC)
[edit] the merge thing
I suppose I'd be in favor of merging in supernet and Provider-based addressing, mostly because those articles are not only poorly written, but inaccurate; having them go away in lieu of the content here seems like a good thing. The subnetwork article also needs an awful lot of work to make it jibe with reality, but I don't think that it'd be correct to merge it in here. --moof 05:33, 4 December 2006 (UTC)
I'm opposed to merging Classless Inter-Domain Routing into Provider-based addressing since CIDR is by far the more commonly used term (of course, this is only from my personal experience). Perhaps doing the opposite and folding Provider-based addressing as a subsection into CIDR would be a better idea? Scraimer 00:08, 16 February 2007 (UTC)
