Template talk:OSIstack

From Wikipedia, the free encyclopedia
Jump to: navigation, search


DHCP is listed as level 2. Shoudn't it be on a higher level. For instance level 7? (talk) 20:39, 7 May 2011 (UTC)


Hi there people,

I see the E1 and SONET/SDH are placed as layer 1 in the OSI model frame. My doubt is that E1 and SONET/SDH also treat the information at the logical level,for example, when processing CRC checks, analysing the time slot 16 (E1 PCM30) and many more in the case of the SONET/SDH. So, in my oppinion, these standards should have a secondary specification, as for example E1 PHY, or SONET PHY, to clarify about which part of the standard we are talking about. The CRC-4 feature, for example, is a feature related as layer 2. As stated in the artcicle "The Data Link Layer provides the functional and procedural means to transfer data between network entities and to detect and possibly correct errors that may occur in the Physical Layer." This is an example of the physical and logical characteristics of these standards (E1 and SONET/SDH).


Carlos Jazbinsek (talk) 15:37, 8 February 2010 (UTC)

The CRC4 is a layer 1 feature. CRC4 doesn't transfer data between network entities; CRC4 appeared in order to avoid FAS simulation on PCM streams and improve quality supervision on the receiver end (no error correction, only detection). Data in time slot 16 is CRC16 checked. User's Access to ISDN layer 2 protocol (Q.921) use CRC16 error detection to ask retransmission (and thus "correct" error). Cardeña. — Preceding unsigned comment added by (talk) 18:45, 22 November 2011 (UTC)

OSI and IP model are not the same[edit]

Dear all,

there is a huge misunderstanding of what the OSI model is. The article on the OSI actually represents this very clearly. Note that the OSI model is a structure that applies (fully or in parts) for any digital communication network. Fitting candidates are ISDN, SDH, ATM, Ethernet, RPR, IP, and so on.

IP lacks its own lower layers and therefore must use the network layer of some other network scheme to have transfer functions that are capable to get an IP packet from one router to the next. Commonly this is today Ethernet, but other were used and are still used as well (ISDN, SDH, ATM). For a clear view all protocols that are capable to provide an end-to-end connection passing some intermediate node (e.g. a cross-connect, a switch, a router, or something similar) should be placed at OSI Layer 3.

The OSI IP model defines a hierarchy of these network layers, and that is accidentally also refereed to as the OSI model, causing a lot of misunderstanding and confusion.

Therefore any figure showing other network protocol's acronyms never refers to the technology independent OSI model defined in ITU-T X.200. (talk) 12:08, 24 May 2010 (UTC) by Sprawl.

Sprawl seems to have a point here: This template does not describe the OSI model but the population of the OSI model with protocols from the TCP/IP protocol stack. I guess that is what he means by "OSI IP model". In any other context, IP, TCP, UDP, and so on, are not part of OSI in any way (not of its model, and not of its protocol stack), and the protocols that are part of its protocol stack, like IS-IS, ES-IS, TP2, RTSE, have somehow not made it into this template. --Pgallert (talk) 14:58, 17 January 2011 (UTC)

Where do we put this?[edit]

Some articles have put it at the top of the page. Others put it after the lead. I assume consistency is good. What's the best place? --Kvng (talk) 16:53, 4 July 2010 (UTC)

Layer 5 vs. Layer 7[edit]

There are huge misunderstandings in this graphic of what Layer 5 to 7 protocols are. The graphic shows protocols such as HTTP and SNMP at layer 7 when they are clearly layer 5 protocols. The corresponding layer 6 for these are HTML/XML for HTTP, and MIB (via OID's) for SNMP. The layer 7 is then your web browser or MIB browser. SIP is shown at layer 5 and 7 while it is clearly a layer 5 protocol. Some of the others like FTP and Telnet don't fit into this definition as they combine layers 5-7.

I agree with the previous poster, any figure showing any network protocol acronyms never refers to the technology independent OSI model defined in ITU-T X.200. —Preceding unsigned comment added by (talk) 04:54, 25 July 2010 (UTC)

Anon is right here. Anyone to else to discuss this? -- I do realise that previous attempts to put that straight have been reverted without discussion. --Pgallert (talk) 14:58, 17 January 2011 (UTC)

Why is the color Blue in the PHY layer?[edit]

Someone should fix the link to Blue or remove it.

SPDY should be Level 7[edit]

According to its wikipedia entry, SPDY is application-level ( http://en.wikipedia.org/wiki/SPDY ). Why is it at level 5? — Preceding unsigned comment added by (talk) 22:38, 30 July 2011 (UTC)

Corrected. -- Tomtefarbror (talk) 16:50, 28 August 2011 (UTC)

PPTP belongs to the data link layer[edit]

PPTP belongs to the "data link layer". You can find in this table OSI_model#Examples and in lots of places in internet. If nobody will object it, I will move PPTP to the data link layer in 7 days.--Sena (talk) 08:33, 15 September 2011 (UTC)

Well, I was wrong. PPTP does not belong to the data link layer. You can check explanation here. --Sena (talk) 16:49, 15 September 2011 (UTC)
Now there is a confusion. In OSI_model#Examples it is still listed in data link layer. Should we put an explanation to discussion this discrepancy? Yenweiliu1 02:14, 28 March 2013 (UTC)

Devices VS Protocols.[edit]

(Copied here from Talk:OSI_model#Two_devices_in_protocol_sidebar.)

In the sidebar, "hub" is listed as Layer 1. While a hub is a device that operates at Layer 1, I don't believe physical devices should be listed on the table. Likewise for Network Switch listed on Layer 2.

If devices are going to be listed, where does it end, pots is listed, should we then list all the devices that pots plugs into(DMS, 5E, Telica, PBX, etc...) I don't think that's the route the article should go, let the Ethernet Hub, Ethernet Switch pages describe there functionality in reguards to the Protocol(s) they run, IE 802.2/802.3/Ethernet II, etc....

— Preceding unsigned comment added by KMurphy111 (talkcontribs) 15:26, July 12, 2012

Good point! The right place to discuss this would be Template_talk:OSIstack, the discussion page of the sidebar. You may want to post you request there. I'll support it. Cheers, --EnOreg (talk) 23:19, 12 July 2012 (UTC)
I disagree. The sidebar may not be complete, but that's not it's purpose here. It only lists a few popular examples for each layer, so one can get an understanding of the model. One could perhaps annotate that it is not complete, since it seems that this is not obvious. --GGShinobi (talk) 11:15, 15 July 2012 (UTC)
My point is that devices are not really part of the OSI Model I have a managed switch here, it runs at all layers, has a web server in it, as well as telnet shouldn't it be up to a page on a device to describe how it operates in reguards to the OSI Model. Protocols however, do map(better) to the OSI model. Having devices in the sidebar will confuse new people trying to learn about OSI and mislead them to think think that devices are part of the model.Kelly Murphy 14:29, 16 July 2012 (UTC)
— Preceding unsigned comment added by KMurphy111 (talkcontribs)
IMHO it is simply inconsistent to mix two example devices with 100 protocols. --EnOreg (talk) 21:29, 17 July 2012 (UTC)
Done. --EnOreg (talk) 19:39, 24 July 2012 (UTC)

X.25 doesn't fit[edit]

Should we even be mentioning X.25 here? It spans layers 1, 2 and 3 and, according to the the X.25 article, is data-linked w/LAPB. AngusCA (talk) 16:50, 20 June 2013 (UTC)

In the industry, X.25 is differentiated even though there is only one spec: X.25 LAPB is the layer 2 part, X.25 PLP (Packet Layer Protocol) is the layer 3 part. LAPD, LAPF, and several other LAP protocols are very similar to LAPB (LAPD and LAPF are subsets of LAPB). LAPD is defined in Q.921 and LAPF is Q.922. IIRC X.25 doesn't specify a physical layer (but I've only seen it on HDLC) — Preceding unsigned comment added by Hank1230071224 (talkcontribs) 10:39, August 5, 2013

Network / Link Layer Muddle[edit]

I am puzzled by the way the various protocols have been classified in these pages. My understanding is as follows.

The OSI Physical Layer = the definition of the physical infrastructure capable of providing an (unreliable) symbol stream between two directly connected endpoints. The symbols are often bits, but can be from any finite alphabet. The key characteristics of the physical layer are (a) that it is between two directly connected endpoints (there is no concept of network address), and it is unreliable. Examples: RS-232, RS-422, 10Base5, 10BaseT, 100BaseTX, V.22, etc. Copper cable, optical cable, wireless carrier signal, smoke signals, etc.

The OSI Link Layer = the layer which turns an unreliable symbol stream connection between two directly connected endpoints into a reliable symbol stream connection between two directly connected endpoints. The key characteristics of the link layer are (a) that it is between two directly connected endpoints (there is no concept of a network address), and it is reliable. Examples: v.41, v.42, v.42bis. PPP, Zmodem, Kermit. Error-correcting code / error-detecting code with retransmission protocol, superimposed on a raw binary channel.

The OSI Network Layer = the layer which turns reliable direct connections between pairs of directly connected endpoints into a network, i.e. a communications medium in which each endpoint can send a packet to another endpoint by relaying its packet to its (directly connected) gateway and providing to the gateway a delivery address on the network. The sender does not have to worry further about the message getting delivered — the network layer provides the necessary routing to deliver the message. Message delivery is not necessarily reliable (delivery is not guaranteed and the sender does not know if it has been delivered). The key characteristics of the network layer are (a) that it is between a sender and an addressee (there is a concept of a network address), and that it is unreliable.

The OSI Transport Layer = the layer which turns an unreliable channel for sending addressed packets into a reliable channel for sending addressed packets. The key characteristics of the transport layer are (a) that it is between a sender and an addressee (there is a concept of a network address), and that it is reliable.

The Internet (IP) and the Ethernet layer are both OSI Network layer protocols, where the IP network layer is layered on top of the Ethernet network layer. The Ethernet layer is not a data link layer.

There is nothing unusual about stacking OSI Layers in a way which doesn't build OSI Layer (n+1) on top of OSI Layer n. PPPoE is an OSI Link Layer (PPP) stacked on top of an OSI Network Layer (Ethernet). ZModem run on a serial line provided by a modem which already provides v.42bis error correction is an OSI Link Layer stacked on top of another OSI Link Layer. ZModem run over ssh is OSI Link Layer run on top of an OSI Application Layer.

To distinguish between the two network layers, the early pioneers of the Internet (who were familiar with the OSI Model) referred to the network layer(s) on top of which the IP protocol operated the network layer, or the network connection layer, and to the new network layer which they were creating, which was connecting existing networks (and network layer protocols) into a bigger network (and more universal network protocol) the inter-network layer.

It is also not the case, as is claimed on many pages in this series that there is no correspondence between the OSI layers and the TCP/IP layers. The layers are clearly defined, and the correspondence is there and it is very clear.

E.g. IP is an OSI Network Layer protocol implemented on top of other OSI Network Layer protocols, such as, e.g. the MAC/Ethernet layer.

I have re-arranged some of the protocols in line with the above definitions, though I think several of them are still in the wrong place.

Tarian.liber (talk) 01:00, 29 August 2013 (UTC)

Reliability is not the distinguishing feature between layers. It may or may not be implemented at any given layer. For multiple-access media such as coax Ethernet, addresses are required at the link layer. Ethernet is most certainly a layer-2 protocol, not layer 3, as you will find in any text book on computer networks.
Please discuss further changes before you make them---in particular in templates that effect many articles. Thanks, --EnOreg (talk) 08:08, 29 August 2013 (UTC)
It would be very helpful if you could please kindly stop reverting my edits and try to resolve this issue on this talk page. As regards to what one will find in "any text book on computer networks", if you are going to appeal to textbooks, please kindly provide a relevant specific reference. I cannot, of course, claim to have read every relevant publication on the subject matter, but I have read enough and am quite familiar with all the main books written by competent authors in this area to be able to say with some confidence that in my carefully considered and definite opinion, I think you will find, if you do actually read any textbooks on computer networks, it quite difficult, very difficult in fact, to find a textbook which states what you claim. I quote, for example, the well-known textbook by K.G.Beauchamp, Computer Communications, Second edition, Chapman and Hall, 1990, p. 208-209. You will find that other good textbooks say exactly the same.
"Data link layer: ... operates to overcome the deficiencies of the physical layer ... It carries out error detection and correction for errors arising during transmission and secures data synchronization. In effect, it converts a simple physical, but possibly unreliable connection, into a tested and error-free connection between two end locations.
Network layer: ... converts the reliable digital connection provided by layer 2 into a multinode network exchange of data."
Which is precisely what I wrote in my edit.
It is incorrect to say that reliability is not a distinguishing feature between layers. Layers 2 and 4 provide reliability on top of layers 1 and 3, each a reliability of a different kind, and providing this reliability is their function and their only function.
It is also not correct to say that addresses are, or can be, required at the link layer. As Beauchamp says, the link layer provides a "tested and error-free connection between two end locations", and does not know about addresses, whereas it is layer 3 that "converts the reliable digital connection [between two points] provided by layer 2 into a multinode network exchange of data". The moment you require addresses to communicate, you are at layer 3 or above.
As for Ethernet, for this same reason, it most certainly and definitively defines a layer 3 protocol, and it is the layer 3 protocol which is common to all the various variants of Ethernet. It is, however, true, that the Ethernet standard provides a whole stack of protocols, going all the way between layer 1 and layer 3 (it supports a large number of physical layer protocols such as 10Base2, 10Base5, 10BaseT, 100BaseTX, etc.). But what is shared by all of them — the top of the stack — lives in layer 3.
Tarian.liber (talk) 17:59, 29 August 2013 (UTC)
Now we have one more protocol to discuss: Wikipedia. In your edit summary you accuse me of edit warring. Well, it takes two to edit war. If you were more reasonable than I you would stop reverting. Moreover, as I wrote in my previous edit summary which you chose to ignore, "controversial changes must be discussed before inclusion." The controversial change is, obviously, your original edit. I'm referring to the Wikipedia:BOLD, revert, discuss cycle: You've been bold in making your original edit, that's fine. I disagreed with your edit and reverted, that's fine. The next step would be to discuss the change until consensus is reached while leaving the article in its original state. Instead you continue to push your controversial change. This violates the protocol.
To the technical questions: I'm not familiar with Beauchamp and it is hard to get hold of a 23-year old book. Does Beauchamp really explicitly say that the link layer "does not know about addresses"? Does he really say, "The moment you require addresses to communicate, you are at layer 3 or above"? I have a hard time imagining that any textbook would be so wrong but it seems to be at least very confusing. What does he say about multiple-access media?
One of the standard textbooks on the subject is Tanenbaum's Computer networks. I'm looking at the 5th edition (2010) where he describes Ethernet in the chapter on "a sublayer of the data link layer called the MAC (Medium Access Control) sublayer." (p. 257) Here's a tutorial from the IETF that should make it clear as well: http://www.ietf.org/edu/documents/82-RoutingBridgingSwitching-Perlman.pdf
Reliability in the networking context usually refers to acknowledging transmission to detect dropped packets. Now I see that you actually mean error correction. That indeed also happens in layer 2. And, being a layer-2 protocol, Ethernet does it, too.
Any other open questions? --EnOreg (talk) 09:12, 30 August 2013 (UTC)
As no further arguments have been brought forward after three days, considering the significant visibility that an error has in a template that many articles include, and referring to WP's BRD cycle mentioned above, I hope you understand that I'm now reverting again. --EnOreg (talk) 09:32, 2 September 2013 (UTC)
Perhaps you have the time to read these pages every day, but I don't. "A whole three days" have gone by without comment from me, therefore you consider the matter closed. I'm afraid have only just seen your comments. I have nothing further to add, as everything is already covered by Beauchamp, a classic reference which is required reading for anyone claiming to be an expert in this field, or even to be worthy of an undergraduate degree in the subject. You have not addressed my (his) points at all. And you have some gall in poo-pooing his book. Perhaps you are of a younger generation who was still wearing his nappies when Beauchamp wrote his book, but this gives you no right to dismiss him so flippantly. My edit was not controversial at all, but was merely making a factual correction to a factual error. I have therefore not been bold at all in making it. You keep re-introducing factual errors. Go find the book and read it, then start talking.
Tarian.liber (talk) 18:26, 2 September 2013 (UTC)
This layer stuff is, for some reason, an emotional topic for editors. I'm reverting Tarian.liber's edits because including IEEE 802.3 as a Network layer technology is clearly wrong as per the content of the IEEE 802.3 article. If you think IEEE 802.3 has it wrong, please go discuss it there before making changes here. If this is not an acceptable temporary resolution, we can bring in some additional editors from Wikipedia:WikiProject Computing/Computer networking task force or WP:RFC to talk about it more. ~KvnG 13:33, 3 September 2013 (UTC)
Well a "model" is just that, a way of thinking about reality. It should not be confused with reality itself. There certainly is a history of muddle through the decades, although just because one book from 23 years ago says several of the protocols that were called "data link layer" at the time did error correction or retransmission, does not prove anything. The arguments about "reliability" are not relevant since each layer can provide reliability over the ones below it, or not, depending on if the designers think it is needed or worth the cost. The quotes above say nothing about where 802.3 goes. Certainly a few of the huge family of 802.3 protocols crept up into what could be called network layer (inter-bridge routing for example). But I would say there is overwhelming consensus that put the vast majority of 802.3 protocols at the physical layer (perhaps some creeping up to data link of course). For that matter, 802.3 is not the same as Ethernet (both are large families). And of course some of us have been doing computer networking since before the OSI model was developed, and worked with some of its key developers, so personal attack are not going to further your case. W Nowicki (talk) 17:50, 3 September 2013 (UTC)
It's not just one book from 23 years ago. It's every book from 23 years ago. And the OSI Layers were defined a long time ago, a long time before IEEE 802.3 and the Internet. They were defined — as this, and every other book, from that period, will confirm, as follows:
Layer 1 - Physical layer - defines physical interfaces (electrical voltage levels, etc., physical connector sizes, etc.) and outputs an unreliable stream between two directly connected endpoints.
Layer 2 - Data link layer - converts what it gets from the layer below into a reliable connection between two directly connected endpoints.
Layer 3 - Network layer - converts what it gets from the layer below into a multipoint network where nodes can send unreliable addressed packets between each other.
Layer 4 - Transmission layer - converts what it gets from the layer below into a multipoint network where nodes can send reliable addressed packets between each other.
This is how every (credible) book of the era when the OSI Layers were defined defines the layers. It is a simple definition, it is an unambiguous definition, it is a clear definition, it is a testable definition, it was deemed to be a useful definition at the time it was made. It does not leave room for philosophizing, and for voting, and for consensus building. This is not social science, it is science science.
This was in the days before WiFi and when most physical interconnect was wires between two endpoints.
You may argue that it's a bad definition, you may argue that it's not a precise definition (after all, what's "reliable"?), but that was OSI's definition. You don't like the layering, you don't find it useful (and indeed, today it may not be as useful as it was back then; to the extent that it was), make your own layering definition, but you can't retrospectively change OSI's, especially not in a way which makes the layering subjective, woolly, and not clearly defined and turns science into matters of opinion and consensus.
I have provided the relevant reference to a reputable book which counts as a reliable source by Wikipedia standards. Or you can go look at any other book from that era. At first the counter-argument is that it's difficult to find the book. Then, that one book doesn't prove anything. Well it's not one book, it's every book, and I have provided a reliable source and you have provided nothing. Not only did you not provide a source for your alternative contention, you have not provided any notion of what the alternative concise, clear, testable and meaningful definition of layers you claim is.
"including IEEE 802.3 as a Network layer technology is clearly wrong as per the content of the IEEE 802.3 article" is a non-sequitur, because another Wikipedia article is not a reliable source for the content of this article, nor does another article saying nonsense justify including nonsense in this article.
We do not need authorities and sources and proof re. whether or not the top of the IEEE 802.3 stack lies in the Network layer or not, since this is decidable by anyone from the first principles of what the layers are. IEEE 802.3 provides a method of sending addressed packets between several nodes on a network, therefore it is a layer 3 technology or above.
I have provided a reliable source, and you have not, so please kindly stop reverting my edits. Thank you.

Tarian.liber (talk) 22:21, 3 September 2013 (UTC)

I gather you think the IEEE 802.3 article is wrong to call itself a data link layer technology but it seems, for some reason, you would prefer to discuss that here not there. There aren't usually a lot of editors watching Template talk pages but I'll see if I can get summon some help over here to help sort this out. ~KvnG 03:24, 4 September 2013 (UTC)
Oh, and one more question, what specifically are the reliable sources you are using to establish your point of view on this? ~KvnG 03:27, 4 September 2013 (UTC)
I am not interested in IEEE 802.3 or that page, or any specific protocol. The editors of that page can choose to be wrong or right, as they want to be. I am interested in the principles of the OSI Stack and in keeping the layer definitions correct. I have given you my reliable source, and I am not expressing any point of view. I have provided the definitions of OSI Stack layers, as generally understood (and I referenced and quoted Beauchamp's book). Are you saying that this (as I briefly and succinctly reproduced above) is not how the layers were defined by the OSI model? If you disagree, how, in your view, does OSI define the layers? Or are you saying there is no definition, and that layering is left to our artistic interpretation? The definition itself is easy enough for anyone to apply. Please explain what you are disagreeing with — the wording of the definition, or how it is being applied?
Tarian.liber (talk) 08:52, 4 September 2013 (UTC)
I would just like to be able to cleanly verify that IEEE 802.3 includes network layer capability. Do you have a reliable source that says this is so or are you making this determination yourself? ~KvnG 14:10, 4 September 2013 (UTC)
It provides the capability for nodes to send addressed packets to other nodes on a multi-node network, does it not?
Tarian.liber (talk) 23:21, 5 September 2013 (UTC)
Only within a broadcast domain. Ethernet addresses are link-layer addresses, as defined in the OSI standard, section --EnOreg (talk) 15:07, 6 September 2013 (UTC)

It is difficult to have a discussion if you ignore counter-arguments. I have made several attempts to explain the Wikiquette but you keep violating it. When you get into a discussion like this you cannot expect others to wait half a week for your reply. Ad hominem attacks do not help your position.

Even university libraries don't carry Beauchamp. Unfortunately you won't answer clarifying questions about what he really says either. The two references I gave, in contrast, are reliable beyond doubt and readily available. You're also wrong about history: the OSI reference model does not predate TCP/IP. But this is irrelevant to the discussion. And again, error detection is not reliability.

Your main argument appears to be that there are no addresses at the data-link layer. Here's what the actual OSI standard has to say (ISO/IEC 7498-1, p.46):

" In connection-less mode, the facilities provided by the Data Link Layer are:
a) data-link-addresses"

It says nothing about "directly connected". What other argument did I miss? --EnOreg (talk) 08:59, 4 September 2013 (UTC)

Beauchamp is not the obscure book you are trying to portray it as, it is readily available, and it is not true that university libraries do not carry it. Perhaps not every university library does, but all the university libraries I have used, a large sample, do. Nor do any other books of that era say anything different. Why are you mis-quoting me and attacking straw men rather than my arguments? I said the OSI model predated IEEE 802.3 and the Internet. This is certainly true. I learned about the OSI model in at least the 1970s, whereas IEEE 802.3 as a standard didn't make an appearance until the 1980s. And the Internet came after IEEE 802.3. There wasn't a single unified dominant network until quite late in the game, I remember in the mid 1980s, UUCP networking was still quite common. I am not sure who you're arguing with re. error detection and reliability, I was merely quoting Beauchamp. However, I am not quite sure how you propose to layer a reliable bitstream service on top of an unreliable bitstream service other than via redundancy (= redundant coding = error detecting/correcting code). And the dates do matter. The OSI layers were defined when they were, the definitions were simple, testable and unambiguous. The original documents — and the books that were published about them — are the most authoritative, not some later "re-interpretation" by people born in an era in which teen-agers define a typewriter as a laptop with a built-in printer, who do not properly understand the context in which those definitions were made. As regards me taking a whole half a week to reply, I'm afraid that's just the pace at which an old fogy like me, who has other things to do than edit Wikipedia, moves along, and you'll just have to learn to be a bit more patient.
Tarian.liber (talk) 23:21, 5 September 2013 (UTC)
We don't have to rely on any textbook representation by simply going back to the actual OSI standard, as I did above.
If you feel I argued against a straw man I apparently didn't understand your arguments. What are they, very concisely?
Both IP and OSI were developed in the late 70s and standardized in the early 80s: IP in 1980 (RFC 760 which was preceded by several IENs) and OSI not until 1984 (ISO 7498). Metcalfe and Bogg's seminal Ethernet paper was published in 1976, the 802.3 draft standard in 1983. What difference does it make?
The point about reliability is merely terminology. Using different terminology makes arguments hard to follow, that's all.
Like ad-hominem attacks, condescension doesn't help your case. --EnOreg (talk) 15:07, 6 September 2013 (UTC)

Tarian, there are two issues here. The first is that this is a template talk page, and that's not where content is being discussed, and not where content decisions are being made. The template collects what the article describes. I don't agree with many of the placements of protocols on particular OSI or TCP/IP layers, but this is not the place to discuss that, and this is not the place to get consensus on content changes.

The second issue is that there are thousands of books about it, and that quite a few of them don't get it quite right. If the definition of the first four layers is in Beauchamp as you quoted it here then this book belongs to that large group; there is almost nothing right in those four layer characterisations, not even the layer names (There's no Transmission Layer). Yes, books are reliable sources but there is no problem providing other books that don't put it that way, as has happened in the article on the OSI model, for instance. Hope this helps, Pgallert (talk) 11:53, 4 September 2013 (UTC)

"Transmission" was my typo, the book correctly uses the term "Transport", as do I, most of the time. I apologize for the use the wrong "Tr..." word. To be fair on Beauchamp, the only thing you should blame him for is the things I actually quoted, in quotes. As regards talk, I agree and agree to discuss matter on other pages, indeed I added my main comment to the main OSI Stack page, and added a copy of it here merely because I edited this page and to give reasons for this edit.
Tarian.liber (talk) 12:26, 4 September 2013 (UTC)
At this point, we're far enough along that I don't think we should move this conversation. Moving it would risk restarting it and these layer discussions go on long enough as it is. ~KvnG 14:10, 4 September 2013 (UTC)

And can we get back to the original issue being discussed. So far the quotes do not explicitly say that IEEE 802.3 protocols are in the network layer. In fact, from sources I see that the OSI model was developed from 1977 to 1979 (totally missing from that article, sigh) while 802.3 did not come out until 1983 (although Ethernet of course was done before OSI), so all three (OSI, IEEE and IP protocol suite) were essentially going on in parallel. As a compromise, I would say it might be safe to say that the IEEE (and IP to a lesser extent) protocols do not fit directly into the OSI model, so they should be removed from this template. In fact, the template is not transcluded in the IEEE 802.3 nor IEEE 802.2 articles, which is another good idea to remove them from this template. And eventually we do need to discuss the edits to the OSI model article too which are a bit misleading. For example, the OSI model (or at least the ISO protocol stack that resulted) had both connection-oriented and connectionless network layers. In connection-oriented networks, virtual circuits were the abstraction, not datagrams, which were associated with connectionless networks only. Many (not all certainly) were from the phone companies after all. W Nowicki (talk) 16:09, 4 September 2013 (UTC)

IEEE 802.3 is certainly a suite of protocols with a wide variety of functions. The notion that the standard provides its own protocol stack and layer definitions is well known and not unusual. However, the standard itself places its significance into the Data Link Layer and the Physical Layer of the OSI model.
Quote: "This standard is organized along architectural lines, emphasizing the large-scale separation of the system into two parts: the Media Access Control (MAC) sublayer of the Data Link Layer and the Physical Layer. These layers are intended to correspond closely to the lowest layers of the ISO/IEC Model for Open Systems Interconnection (see Figure 1–1). (See ISO/IEC 7498-1:1994.1) The Logical Link Control (LLC) sublayer and MAC sublayer together encompass the functions intended for the Data Link Layer as defined in the OSI model."
This paragraph references a figure which unambiguously illustrates the intended operation layers of this standard. This template reflects the definitions of the standard. To try to reinterpret the intent of the standard by some old obscure text book is rather absurd, when dozens of more recent and well-known secondary sources confirm the standards-based interpretation. Case closed. Kbrose (talk) 00:26, 5 September 2013 (UTC)
The notion that a standard provides its own protocol stack and layer definitions is not unusual, so long as it uses its own layering scheme. If it uses the OSI Layering scheme, then, if the standard wants to be correct, it has no choice but to apply the OSI Layer definitions. Of course, the writers of any document can choose to do whatever they want, but that does not make it correct. Since this is a page about "OSI stack" and not "IEEE 802.3", I submit that it's the OSI stack definitions in their original form (not a form re-defined by someone born in an era in which HTTP is the physical layer, and nothing else exists), and not derivative standards and documents, which are authoritative.
Tarian.liber (talk) 23:21, 5 September 2013 (UTC)
Thank you, W Nowicki, for providing the exact dates for IEEE 802.3 and the OSI model. I was having a debate above with someone else about which came sooner. I knew OSI dated back to the 1970s and IEEE 802.3 to the 1980s, but I did not have the exact years, and those sound about right. I'm sure you have checked them to make sure they are correct. And yes of course Ethernet pre-dated its standardisation as IEEE 802.3. Re. your suggestion to say that Ethernet and IP do not fit exactly into the OSI model, and removing them from this template, my thoughts are as follows. I think removing "IEEE 802.3" (and some other standards too) from this template is a very good idea, because "IEEE 802.3" does not fit into any single OSI layer, parts of various sub-standards of IEEE 802.3 belong to various of the bottom 3 OSI layers, and elaborating on the detail of what fits where goes beyond the scope of such a template summary page. On the other hand, 100BaseTX etc. clearly belong to the physical layer (OSI Layer 1), so in my view they could stay, but if that's going to cause upheaval, I can't see any harm being done by deleting them too. I don't think it's the case, though, that IEEE 802.3 does not fit into the OSI Layer model. The OSI Layer model is precise, well and clearly defined, and it's easy to decide which part of which IEEE 802.3 standard belongs into which OSI Layer. I think this is the underlying cause of the current disagreement, in that I am taking the stance (as did the early publications about OSI) that OSI layering is simple and clearly defined, and that to decide what belongs where, you look at the definitions and apply them, whereas opponents of this position are taking the view that it's really a subjective matter, that OSI Layers don't have any clear definition, and that it's up to the writer of a communications standard to decide where in the OSI model he fits bits of his standards, and that whatever he decides and however he decides it, that's then authoritative, even if he labels networking functionality "physical layer" or "link layer" or whatever. I think this is the crux of the matter.
Tarian.liber (talk) 23:21, 5 September 2013 (UTC)
I don't think anybody here has argued that Ethernet does not cover OSI layers 1 and 2 and I don't see any reason to remove them from there. This discussion is about whether Ethernet provides layer-3 functionality.
I don't think anybody here has claimed that the classification of protocols into the OSI layers is "really a subjective matter". My understanding has been that we disagree on the OSI layer definition. Looking into the actual standard may resolve this problem. --EnOreg (talk) 15:07, 6 September 2013 (UTC)

Sorry, I'm getting a bit forgetful these days and I forgot what my own point about the timeline was, and now I remembered. To those who claim that Ethernet does not define a layer which belongs to OSI Layer 3 (and that the relative timeline of coming into being of OSI and the Internet does not matter), I ask this question: before the Internet came along, which stitched together all the other networks, and introduced an 'inter'-networking network layer on top of the network layer of other networks, and before TCP/IP became standardized, people were already using Ethernet for their networking. So which part of their (our) networking stack formed OSI Layer 3 in those days then? Just answer this question for me, please.

Tarian.liber (talk) 02:03, 6 September 2013 (UTC)

Does this go back to the claim that there are no addresses on layer 2, which the above quote from the standard clearly refutes? How do you define the borders of a network for the purpose of this question? A broadcast domain or beyond that? --EnOreg (talk) 15:07, 6 September 2013 (UTC)


I see that TLS/SSL is shown in this template as a layer 5 protocol. I think this is misleading. TLS/SSL do not operate at layer 5. In the Transport_Layer_Security article there is a paragraph stating that TLS is initiated at layer 5 and operates at layer 6 and there is an infobox statig that TLS operates at layer 7 - so we are not being consistent. I think the reality is that TLS does not really fall nicely into any layer of the model and I would suggest its clearer to just remove it from this template altogether. Nicholsr (talk) 17:44, 3 January 2014 (UTC)

I support including only protocols whose layer membership is well established and agreed upon by sources. Trying to be more comprehensive is unproductive and potentially introduces original research. ~KvnG 00:07, 7 January 2014 (UTC)
Thank you, I have removed it Nicholsr (talk) 13:33, 15 January 2014 (UTC)