Talk:OSI model

From Wikipedia, the free encyclopedia

	This article is part of WikiProject Computer networking, an attempt to build a comprehensive and detailed guide to Computer networking on Wikipedia. If you would like to participate, you can edit the article attached to this page, or visit the project page, where you can join the project and/or contribute to the discussion.	Computer networking Portal
B	This article has been rated as B-Class on the assessment scale.
High	This article is on a subject of high priority within Computer networking for inclusion in Wikipedia 1.0.
Computer networking To-do: edit  Computer networking Collaboration of the week: Wikipedia:Computer networking Collaboration of the week/current Template:Cnbox contents

This article is within the scope of WikiProject Telecommunications, an attempt to build a comprehensive and detailed guide to telecommunications on Wikipedia. If you would like to participate, you can edit the article attached to this page, or visit the project page, where you can join the project as a "full time member" and/or contribute to the discussion.

Contents 1 Copies content 2 OSI Graphic - Osi model com.jpg 3 Other ideas 4 Need help 5 Problems with the OSI model 6 Layer 0 7 MPLS 8 Colors 9 some of the mnemonics are wrong (grafitti) 10 methinks a few mnemonics are misogynistic 11 Making sense of the OSI model 12 Removing the mnemonic devices 13 X.25 and OSI 14 Links to seven layers 15 Layer 2.5 16 Mnemonics 17 Capitalization 18 Security services on layer 6 (encryption). 19 HTTP Layer 7 ??? 20 RIP is a routing protocol 21 Images 22 rocks! 23 Freely-downloadable ISO 7498-1 OSI Reference Model document 24 Proposed merge with "Open Systems Interconnection"

[edit] Copies content

A search on Google shows that this page has paragraphs found multiple times accross the internet (right down to the space in 'a re' in the last paragraph). Whether this means its in the public domain, or that they are all stealing from the same source I am unsure. If its the former, something needs to be cited. - Eean

Much of it seems to be from http://www.btconline.net/~derrellh/OTI/OTI_Classes/CIS140/OSI_model/OSI%20model.htm. It seems highly doubtful that this is public domain. I've removed the content for now. --Stephen Gilbert

[edit] OSI Graphic - Osi model com.jpg

Osi model com.jpg

I don't think this graphic is appropriate for an encylopedia. 1. The reference to "The Internet" is not appropriate. The internet is not always network (ie it can be GSM, non-internet TCP/IP nets etc) 2. The figures of two human faces do not add to the information content and can infact be misleading (ie endpoints are not always, human). 3. This images does not bring anything new or significantly aid in depicting the concept, especialls since "Rm-osi parallel.png" and 2 two tables included in the article.

Recommended Action: REMOVE Image "Osi_model_com.jpg" from the article & restore formatting.

Note to Markolinsky: It seems you have spend quite a bit of effert on this drawing and it looks good, but I don't think the graphic style and content is not appropriate in this article. In the future, it is probably best to discuss in the talk page before doing changes.

--202.161.20.46 08:42, 27 September 2006 (UTC)

Removed from article. As said above - Internet usually just works on network layer. --Nux (talk)

[edit] Other ideas

from user talk:MyRedDice

Here's my thinking:

First on OSI you'll notice the 7-long table with some examples of things at each of the 7 OSI levels.

You'll notice ethernet is on level 2 and 10BASE-T is on level one.

goto 10BASE-T and you'll notice that its said to be a "varient of ethernet",

PLEASE help me clarify how a variant of ethernet is on a DIFFERENT OSI level than it.

(see, i think that OSI categorization is just a loosely thrown together piece of ill-defined cruft) Anyway, maybe you can educate me

---

I think I can help. Ethernet is level two. But it depends upon level one: the physical layer. Thus you can run ethernet either on 10BASE-T cables, or on 10BASE2 cables, or on [[please respond to user_talk:hfastedge so that I dont have to constantly look at recent changes.

anyway, "you're just changing which variant of the physical layer it works on" that still doesnt address how "10BASE-T is a variant of ethernet" as mentioned in 10BASE-T.

Have replied there - see new version of 10BASE-T :) Martin

You almost got it right but you responded to user page instead of user_talk:hfastedge . ONLY when u respond to the talk page do you get a little * eg: Hfastedge (Talk*) allowing you to globally be aware of new talks. Finally, you've clearly introduced data within the article that is conflicting (within the confines of the article). I'm going to put the article on watch, and see if anyone interested finally (and solidly) gets it "right".

The confusion here is that Ethernet is both layer 1 and 2. 10BASE-T is just one implementation of layer 1 for Ethernet. It isn't a "variant", it's a specific implementation. --AMillar

Yes. I've added some stuff about that in the section on layer 1. At layer 1, there are a number of physical layers of Ethernet, from the old 10BASE5 to shiny modern 10 Gb Ethernet. At layer 2, they all use the same 14-byte MAC layer header, with optional 802.2 LLC header. Guy Harris 02:06, 31 January 2006 (UTC)

[edit] Need help

i need some help, i gotta learn TCP/IP because i chose it as my written or studied language. this site is UNBELEVIABLY HELPFUL! but i need all the help i can get

just post, my email is suffering problems so i cant check it. :( :P -- Ck12233

[edit] Problems with the OSI model

I'm thinking that maybe I should add some of my comments at the end of Talk:Internet_protocol_suite#Confusing layers (about how the OSI model really divides things up in functional layers, and you shouldn't expect the actual dependency maps of real protocol suites to follow it slavishly) to this page; that will prevent a lot of confusion as to e.g. why BGP, which uses TCP, is at the network/internet layer.... Noel 12:40, 18 Sep 2004 (UTC)

BGP is an application (Layer7) that uses a TCP/IP session between two "speakers" The fact that it exchanges routing information is not relevant for Layer3. The given that those "speakers" live in devices that are called Routers, does not degrade BGP from Layer7 to Layer3. Similar to Telnet/SSH, FTP to a router. Would Telnet to a Switch suddenly make Telnet a Layer2 function? I see no problems with the OSI model, only human errors in interpretation.

[edit] Layer 0

Sometimes a layer 0 is introduced to refer to the physical medium the data is transported accros, eg. the copper or glass cabling itself. This is because strictly speaking the cabling does not belong to the osi model itself. [JP]

[edit] MPLS

MPLS is considerd to be a switching techonology, i.e. layer 2. However it runs on other layer 2 technologies such as Ethernet or ATM, then why is it not considerd layer 3. Then again is what defines a layer 2 Protocol one that specifies the next hop in the path while layer 3 specifies the final destination.

Possibily could MPLS be considerd a sub-layer of Layer 2, so if MPLS ran over Ethernet there would be 3 sub-layers LLC layer, MAC layer and MPLS layer. -Vec 19 April 2005

Yet another perfect example of why the ISO model sucks for explaining anything above framing (e.g. HDLC) and below transports (such as TCP). In a model which had separate "internetwork" and "network" layers, MPLS would be the lowest sublayer of the "internetwork" layer (since it is intended to be able to forward traffic across multiple networks, which may be different kinds of networks). Noel (talk) 18:07, 14 September 2005 (UTC)

Furthermore this is why some call MPLS a "Layer 2.5" because it's both above layer 2 and below layer 3. Cburnett 04:28, 2 June 2006 (UTC)

MPLS is a Layer3 function, without doubt. Not layer2, nor layer2.5 It modifies layer3 headers. The fact that it does it in Layer2 hardware (the switch) does not mean it will work in ANY switch. The switch that can act on MPLS fields in packets is relying on the routing info from routers at the edge of the MPLS autonomous system.

[edit] Colors

The colors in the OSI model image (Image:Osi-model-jb.png) on this page could need a change, since they are bad from the perspective of someone with any common color deficiency. Especially colors in the "Transport" and "Network" fields are hard to tell apart. Would fix it myself if it wasn't that I have both a green and red deficiency and would probably just mess it up. There's a chart with safe colors here and some more info here. -Manwal 23 June 2005

[edit] some of the mnemonics are wrong (grafitti)

I am not changing it though, cause I am starting a new job tomorrow and I have to read some suff to feel ready. —The preceding unsigned comment was added by 64.231.152.31 (talk • contribs) 03:28, 3 August 2005.

These were long since removed, the new ones work. --67.183.217.186 09:49, 9 May 2006 (UTC)

[edit] methinks a few mnemonics are misogynistic

Guys, can we consider deleting references to dicks and pussies? It really detracts from an otherwise useful page. —The preceding unsigned comment was added by 216.184.31.184 (talk • contribs) 23:26, 3 August 2005.

Wikipedia is not censored. However, these were long since removed, and the new ones were bland enough for a Dummies guide (and probably somewhat less memorable, if more verifiable). --67.183.217.186 09:49, 9 May 2006 (UTC)

[edit] Making sense of the OSI model

As I studied the OSI model, I was quite confused, and just didn't get the idea right. Eg, I tried also to map OSI on other protocols, and it never seems to work out. As such, for a long time I saw OSI more or less as an academic toy. It took quite some time before I finally got the sense of it, something I wanna share with you all.

First of all I think the OSI concepts are much more useful then the implementation of the model itself. What are those concepts? Well there is communication, there is layers and there is language. These work in close harmony. The essence here is that each layer communicates with the same type of layer at the other site of the 'communication line'. Let me stretch this: two layers of the same type are talking to each other in the same 'language'. You can see this in almost every picture, but it is never really emphasised. What is the use of this concept? Well it simply helps you to focus on the function of a layer: what does it do, or when designing: what does it have to do.

Next concept is that in the OSI model, except for the physical layer, all layers are separated, and thus actually can not communicate with each other directly. From the model it is obvious that layers are talking too the next lower level (through the so called Service Access Point or SAP). The essence here is that the layer has some means to talk to his counterpart. In other words it tells you how a layer can talk to its counterpart.

Last concept is language. I use the concept of language in the sense of a set of words related to the same domain (for the intended usage database and road don't belong to the same language; road and car generally will). To define a language I ask myself 'what words do I need', 'is the language complete', 'is the language consistent', 'are there any ambiguities'? In every day sense it is sort of pragmatic, semi-formal tool. In the OSI model language is not explicitly defined as a concept, but it's left implicitly. I'm not going to give a precise definition, however you can see as a set of words. These words should be the only words you use to describe a function. And yes: in this way there are many, many 'languages'.

Summary: layers on the same level are virtually functionally communicating to each other in the same language ('horizontally') by really communicating technically ('vertically') to the next lower level layer. These two forms of communication are more or less independant, and to keep things simple you should always focus on one of them at a time.

To me the OSI model is a generic model which solves probably most or all network problems. For my day-to-day work, this is not very useful. So how do I use this concepts? Well, as simple and as obvious as they seem, I use them a lot. First I used them to understand network protocols. It especially helps me to keep out of the OSI-mapping discussion (try to map tcp/ip on these concepts ;-), and it helps me to explain network protocols to a lot of people with ease. Furthermore, if I have to do some work on communication I always ask myself: what layer(s) are we talking about, what language is used and how do they communicate with each other. And yes: this sounds simple, but in my experience 4 or 5 out of 10 don't grasp the idea and take much longer time then needed (if they succeed at all) to complete a job. And this is for network people.

And now the fun part. You can use the same concepts in application development: every time you're working with interfaces, you're talking about communication. The obvious usage is of course for interfaces between processes. Especially during design you can get much clearer discussions. If you 'know where you are' in a protocol stack you can much more easily focus then when several concepts are mingled. A nice example is the man/machine interface. From the above you can conclude that, functionally, only layers who 'speak the same language' can talk to each other. A second issue is that each application has its own 'language'. Working with a financial application is something completely different from working with a video-editor. Each application has its own set of functions, concepts etc. and the human to work with it has to speak the same language. From a design point of view I've turned this around: besides using use-cases etc, in a pragmatic way I formalise the language that the intended users use. Interactively I ask them what they need from an application and I teach them how to talk to a computer. Crucial here is that the language which is developed here is actually their own language, but semi-formalised, and as such it is a language which can be understood by both layers. —The preceding unsigned comment was added by 213.84.162.175 (talk • contribs) 09:02, 21 August 2005.

While the "meta-model" (the use of layers, etc) is fine, the actual ISO 7-layer model is totally useless and confusing for describing everything below transports (such as TCP) and above framing (e.g. HDLC). Cramming X.25, MAC, ARP, IP, ICMP, etc all into one layer is worse than useless. However, the model is widely known, and for some reason (terminal brain damage?) people keep using it. If it had just one extra layer ("internetwork"), it would be a much better model for use with current real-world networks. Noel (talk) 18:07, 14 September 2005 (UTC)

[edit] Removing the mnemonic devices

Removing the mnemonic devices was a bad idea. I came here looking for these (which I'd bookmarked) and found them gone. I dont believe that just because one person finds them hard to remember that they aren't useful.

I agree59.144.250.72 14:53, 23 August 2006 (UTC)

Please see the other discussion of the mnemonics on this talk page. Whether they are "useful" is not the concern. The encyclopedia is for facts. If you want useful tricks for learning things, try reading a Wikibook? ~ Booya ^Bazooka 15:21, 23 August 2006 (UTC)

[edit] X.25 and OSI

As far as I have been able to tell, not all of X.25 is used by the OSI suite (X.25 is a CCITT standard), although some of the protocols (HDLC?) do seem to have been used for OSI. Perhaps X.25 doesn't belong in the OSI column in the table of examples? StuartBrady

[edit] Links to seven layers

The seven layers are described on their own pages, e.g. Application layer.

May I suggest someone adds links from the OSI model article to these pages? Maybe in the section "Description of layers". -- Felix Wiemann 13:59, 14 January 2006 (UTC)

The links are already there — the first link in each subsection links to the article for that layer. One of the links was broken, but I've fixed it now. StuartBrady 19:14, 14 January 2006 (UTC)

[edit] Layer 2.5

I would like to question the usefullness of discussing a "layer 2.5" in this article. It is not part of the OSI model, nor were any of the examples of a 'layer 2.5' protocol designed for the OSI protocol stack. If this *has* to be mentioned, then this should be mentioned after a definition of the formal layers, perhaps even under an 'Informal Layers' header. -- Ryanfantastic 11:01, 31 March 2006 (UTC)

Disclaimer: I added it to begin with. That said, I see no problem in its location (which is why I put it there). The "usefulness" of layer 2.5 is that — seemingly like all communication stacks/models — it points out the need for something between layers 2 & 3. I also don't see the need to reside it to an "others" category: it clearly states that it's not apart of the model and it flows just fine to me. Cburnett 22:45, 31 March 2006 (UTC)

It currently states that MPLS operates on packets (layer 2)... should that not be (layer 3), or frames (layer 2) Thedarxide 09:42, 22 May 2006 (UTC)

[edit] Mnemonics

I have deleted all the mnemonics because how many are being made up. WP:V says the burden of evidence (sources) is on the poster. See here for the mnemonics deleted. Cburnett 05:03, 6 April 2006 (UTC)

You're going to wait for an official ISO standard mnemonic? 88.105.122.117 09:37, 16 April 2006 (UTC)

The editors of CCNA For Dummies supported a few, those are now added and cited as an external link. --67.183.217.186 09:49, 9 May 2006 (UTC)

Excellent. Any source was exactly what I was looking for. And, no, I'm not waiting for an official mnemonic (it'll never come). Cburnett 00:00, 10 May 2006 (UTC)

I deleted the mnemonics again. Just because you have a citation for them doesn't make them valid for inclusion; they're just some things that a For Dummies author made up. If you want to help people remember the facts given in this article, I suggest writing a Wikibook on the subject. For the encyclopedia, please, just the facts. ~ Booya ^Bazooka 05:29, 9 August 2006 (UTC)

[edit] Capitalization

I noticed in this article the names of the layers begin with a capital letter, however in the 7 articles on the layers themselves they are in lowercase (most of the time). This needs to be rectified.

[edit] Security services on layer 6 (encryption).

I was doing a research project on cryptography and in a certain text, (Mitchell, Chris J, Users Guide to Cryptography and Standards, Artech House inc, 2005 [p 32-33]) the authors state the following

"No security services can be provided on in layer 5 or layer 6, although layer 6 may contain facilities to support the provision of services at layer 7."

I must say that I agree with the text, that security services such as encryption are not handled in the functions of presentation layer. It is my personal view that the 6th layer serves the function of interpreter rather then decipherer and as being analogous to the interpreter should not have a hand in the content of the message, or data to be conveyed. Therefore I believe that the functions of (en/de)-cyperment should be excluded from this layer as it is seen in the word encryption.

Magus 05:19, 20 April 2006 (UTC)

[edit] HTTP Layer 7 ???

My view of the layers here is different: TCP Layer 4, HTTP: Layer 5 - It provides a session going beyond one TCP connection (i.e. session cookies etc..) HTML: Layer 6 (presentation) it provides means for prensenting a web page, Layr 7; Appication: IE/streaming applications/IPC etc... —The preceding unsigned comment was added by 67.183.217.186 (talk • contribs) .

I completely agree. HTTP is a session layer protocol; same goes for the FTP protocol. I suspect the confusion here started long ago when things like FTP were listed at the application layer/layer 1. In fact, this was actually somewhat valid: the FTP application (which on most *NIX boxes is called simply "ftp") could perhaps be considered layer one (it isn't a protocol, but the OSI model doesn't only describe protocols), but regardless the ftp protocol belongs at the session layer.

As another example, consider SMTP, which is also a session layer protocol. The article shows MIME as being at layer 6 (which I agree with). MIME encoded messages are transported over SMTP, but SMTP belongs below MIME in the model. But the article shows SMTP above MIME in layer 1/application layer. Iambk 19:53, 22 November 2006 (UTC)

The OSI Model is a protocol stack, I.E. is not a protocol. Another thing to consider: how would your stack look with HTTPS instead of HTTP? Ryanfantastic 18:18, 9 May 2006 (UTC)

First, TCP/IP model is not the same as OSI. The comparison is done to make a comparison: that's it. There is no presentation layer nor a session layer as it applies to TCP/IP and its protocol.s

Secondly, HTML is not a protocol. It's a document/file. Period.

Lastly, I suppose you could, however, make an argument about XML-RPC and SOAP being protocols and can find a place in the stack. Cburnett 23:47, 9 May 2006 (UTC)

(I was the one initially started this section) HTML is a document, that does not mean it cannot be a protocol… however this is not the issue here. For example all Web services which use XML over HTTP, this for sure have more layers than 7…

Also, the statement …”TCP/IP is not same as OSI”… TCP (over IP) is a protocol just as any other protocol, there is not reason to take it out of the OSI model. The OSI model talks about protocol layers (not necessarily 7), for sure there are many layers of protocols on top of TCP/IP which makes it 100% compatible to the OSI model. For example, in MMS: MM4 is a protocol, over SMTP, over TCP/IP – see 3GPP specs. I can give you a lot of protocols very real on top of the “allegedly layer 7” in this article.

Ranc 08:00, 17 May 2006 (UTC)

An HTML file is a file that contains data and a basic set of instructions on how to display the data (markup). You'd have to really bastardize the meaning of "protocol" to make a fixed-content file be considered a protocol. HTML is no more of a protocol than a term paper: it contains data with markup instructions. Is LaTeX a protocol? If you answered yes then you really need to hit the copious amounts of texts on that very subject.

TCP/IP is not the same as OSI. In fact, the OSI model was created after the TCP/IP stack was created. OSI is typically used to explain the TCP/IP suite but that does not mean TCP/IP is OSI or that there must be a one-to-one relationship between them.

What you are completely missing is that the Internet protocol suite#Layers in the Internet protocol suite stack only defines four layers. Everything above TCP/UDP is in the application layer. There is absolutely nothing saying that the application layer cannot have multiple layers to it but they are all still in the application layer.

At the end of the day you have to realize that the OSI model is a model. Models change. Models don't fit all scenarios. Models aren't perfect. Cburnett 04:26, 2 June 2006 (UTC)

The mnemonic "A Perfect Student Needs To Drink Port" does not identify the correct letters for network and transport layers.

cDima 21:13, 13 June 2006 (UTC)

[edit] RIP is a routing protocol

I just reverted someone moving RIP from layer 3 to layer 7.

It's a routing protocol. I suspect that the mover had a different RIP in mind.

There - just wanted to have it on the talk page too. --Alvestrand 20:53, 31 August 2006 (UTC)

[edit] Images

At least one of these images needs to be removed, since they're so similar. I would want to remove them both, though - do the pictures tell us anything that our text tables don't? ~ Booya ^Bazooka 15:46, 28 September 2006 (UTC)

I agree they both should be removed. See my comments re 2nd pic above Talk:OSI_model#OSI_Graphic_-_Osi_model_com.jpg --202.161.20.46 08:11, 29 September 2006 (UTC)

I agree - both should be removed. Your comments about the second picture are spot on. The other picture is better but I don't like the bidirectional arrows between the layers as I believe they are unnecessary and confusing. If those arrows were removed or at least explained then I might feel otherwise. --ElKevbo 15:47, 29 September 2006 (UTC)

I think the image should be removed. It is beyond the scope of illustrating the OSI Model. The OSI Model is a logical model. I appreciate the efforts of the author. I think a solution is 1) remove current image 2) discuss a future image on the discussion page 3) make new image from input. Joneboi 06:00, 19 November 2006 (UTC)

[edit] rocks!

"<mere_raili> čekas norisas yra nulinis OSI layeris"!

[edit] Freely-downloadable ISO 7498-1 OSI Reference Model document

Is it correct, that the link http://standards.iso.org/ittf/PubliclyAvailableStandards/s020269_ISO_IEC_7498-1_1994(E).zip is pointing to a Zip-File containing a PDF, that itself contains a scan of the ISO-Specs which are under Copyright Protection? Please have a look at that. --80.131.151.219 21:57, 24 October 2006 (UTC)

It is, indeed, true that the ISO are allowing some standards, including ISO 7498-1, to be downloaded for free from the page at http://standards.iso.org/ittf/PubliclyAvailableStandards/. That page says:

ISO Copyright for the freely available standards

The following standards are made freely available for standardization purposes. They are protected by copyright and therefore and unless otherwise specified, no part of these publications may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, microfilm, scanning, reproduction in whole or in part to another Internet site, without permission in writing from ISO. Requests should be addressed to the ISO Central Secretariat.

The documents you are about to download are a single-user, non-revisable Adobe Acrobat PDF file, to store on your personal computer. You may print out and retain one printed copy of the PDF file. This printed copy is fully protected by national and international copyright laws, and may not be photocopied or reproduced in any form. Under no circumstances may it be resold.

I don't know why the particular standards offered from that page were chosen. Guy Harris 19:05, 22 November 2006 (UTC)

[edit] Proposed merge with "Open Systems Interconnection"

I'd say "no" - Open Systems Interconnection discusses the overall OSI project, while this page discusses a specific item, the OSI networking model, that the Open Systems Interconnection page claims antedated the project. Guy Harris 19:18, 22 November 2006 (UTC)

Support for not-merge. As per Guy Harris. Visor 14:53, 30 November 2006 (UTC)