Talk:OSI model

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

IP/OSI confusion in layer box[edit]

The box of networking layers here makes the mistake of conflating TCP/IP network protocols with the OSI layers. This is wrong; the two are not connected, not even remotely.

I realize that many educational institutions like to use the OSI model because it's such a neat network model and it's so nice to explain things, but it's 'wrong' to assume that the IP network protocol is situated in layer 4 of the OSI model, or that the SIP protocol is situated in layer 7 of the OSI model. They are not. The IP protocol is found in the network layer of the TCP/IP "model", and the SIP protocol is found in the applicatoin layer of the TCP/IP "model".

If you're going to show the OSI model, the protocols on OSI protocols should be used instead. — Preceding unsigned comment added by (talk) 23:03, 19 April 2013 (UTC)

The OSI model does not only apply to the OSI protocols as the networking literature attests. --EnOreg (talk) 15:26, 21 April 2013 (UTC)
There are no "OSI protocols." The OSI model doesn't define any protocols. By definition, protocols such as TCP/IP fall within the scope of the OSI model. The official documentation says:
"The purpose of this Reference Model of Open Systems Interconnection is to provide a common basis for the coordination of standards development for the purpose of systems interconnection, while allowing existing standards to be placed into perspective within the overall Reference Model."
The tasks generally performed in the Internet layer of the TCP/IP model fall within the general description of the Network layer of the OSI model. Rsduhamel (talk) 23:44, 26 January 2015 (UTC)

Examples wrong[edit]

the table in Examples gives an entry for "Layer 2" X "TCP/IP protocols". This is wrong. Layer 2 protocols, like PPP, is at a level where IP doesn't exist (IP is on top of L2).

Zenkutsu (talk) 18:42, 9 February 2015 (UTC)

Also things like HTTP and HTML are backwards. HTML is transferred over HTTP, so HTML should be layer 7, not layer 6, and HTTP should be layer 6 not layer 7. The same goes for the other 6/7 layer examples.

Majenko (talk) 23:58, 14 December 2015 (UTC)

Defining a standard[edit]

The article says in several places that the OSI model defines a standard. The official documentation specifically says it does not define a standard. Rsduhamel (talk) 16:46, 19 January 2015 (UTC)

But ISO 7498 says right on its front cover that it's a standard. Please explain. --Wtshymanski (talk) 22:07, 26 January 2015 (UTC)
The cover page of the ISO version and page 1 of both versions are titled "International Standard" (from page 1 on, both the ISO and ITU versions are identical). However, the body of the document is clear that the document does not set a standard. Specifically, it says:
"It is not the intent of this Reference Model either to serve as an implementation specification, or to be a basis for appraising the conformance of actual implementations, or to provide a sufficient level of detail to define precisely the services and protocols of the interconnection architecture. Rather, this Reference Model provides a conceptual and functional framework which allows international teams of experts to work productively and independently on the development of standards for each layer of the Reference Model for OSI."
Although it does not say "this is not a standard" it is described as a reference model and framework to allow the development of standards. Perhaps I am being imprecise but I have seen a lot written on the Internet and in textbooks describing the OSI model as a standard that must be complied with. I've also seen it described as an attempt by bubble-headed European bureaucrats to shove a restrictive, complicated and unworkable standard down the throats of developers. It is no such thing. It is just framework describing what tasks may or may not be done by networking applications. Developers don't necessarily follow the OSI model. The OSI model follows what developers generally do. Rsduhamel (talk) 23:29, 26 January 2015 (UTC)
This is really moot, because the OSI model is really used nowhere as a development model, at least not above Layer 2. The lack of usable protocols in the late 80s, early 90s, made it obsolete quickly and TCP/IP has reigned, despite its shortcomings. I don't see a strict contraction in a reference model being a standard. It certainly is a standard way of teaching network application architecture. Developers use the Internet Protocol Suite (TCP/IP) as their model. Anything the standards organizations deal with is codified eventually as a standard, seems that's the only way to document their work. Kbrose (talk) 14:04, 27 January 2015 (UTC)
Call it a standard if you will. There is no point in arguing semantics when the article has much more serious issues. Someone who has actually read the specification needs to give it some serious attention, if it's worth the effort. Rsduhamel (talk) 02:50, 3 February 2015 (UTC)
The whole problem is, even if there is a specification, in a general article such as this one it is sort of futile to attempt to rewrite it. People can go to the spec for an official read. OSI is theoretical not practical. It is not useful to explain it from the detail of an implementation perspective because it is never implemented -- the world has chosen the TCP/IP five layer model, and anyone reading Wikipedia to learn about networking should not waste a lot of time reading about OSI (They will have plenty of time to do this if they ever try to hack an OSI implementation such as HL7.). Therefore, this article should be dramatically simplified.
It is misleading because this page is referenced from pages that discuss very important practical network topics.Squeeky Longhair (talk) 17:57, 9 February 2018 (UTC)

IPoE redirects here. WHY?[edit]

I was looking for info on I(ternet)P(rotocol)o(ver)E(thernet), and got this page. But this page says absolutely NOTHING about IPoE. So where should a seeker after knowledge go? (talk) 03:35, 10 August 2015 (UTC)

Perhaps somebody could restart the IPoE page with this info from Mark Bernstein of Juniper Networks. (talk) 03:57, 10 August 2015 (UTC)

Image formats and styling languages in layer 6?[edit]

In the table in section "Description of OSI layers" there are some interesting examples for layer 6: HTML, CSS, GIF.
How so? Can somebody explain? What does an image format have to do with the presentation layer of the OSI model? And HTML/CSS? This looks like it was mistaken for "visual presentation of a web page".
Similarly SQL is a weird choice for layer 5 because that's a programming language, not a protocol. — Preceding unsigned comment added by Niko IstDerNameImmerNochZuÄhnlich (talkcontribs) 01:35, 5 October 2015 (UTC)

It seems to me that the examples in the upper layers are all completely mixed up. HTML should be a higher layer than HTTP since HTML is transferred using the HTTP protocol over a TCP socket using IP. The 6/7 layer examples are all backwards.

Majenko (talk) 00:00, 15 December 2015 (UTC)

Layer 5 (Session) examples incorrect?[edit]

Numerous text-books and online sources use HTTP; FTP; RDP; SSH; Telnet as examples for Layer 7 (Application).

Even other wiki pages like List of network protocols (OSI model) and Application Layer — Preceding unsigned comment added by Beanaroo (talkcontribs) 09:48, 23 March 2016 (UTC)

Every one of the examples you have listed is an IP model protocol. To quote ISO/IEC 7498-1, which defines the OSI model, "This reference model provides a common basis for the coordination of standards development"; the development of HTTP, FTP, RDP, SSH, and Telnet was NOT coordinated b the OSI model, and did NOT use the OSI model as a basis; instead, the 5-layer IP model model was used when developing these protocols. In IP model terms, layer 4 is the "internetwork" layer, and layer 5 is the "application" layer. Application protocols designed using the IP model as a basis that need session management have to do it built-in, because using the IP model as a basis leads to not having a separate session management protocol. On the other hand, application protocols designed using the OSI model as a basis can outsource it to a separate protocol. Because of this, when using the OSI model to describe IP model protocols, we can say that a protocol spans multiple layers; different parts of the protocol fill the roles in different layers. ~ LukeShu (talk) 07:43, 8 December 2016 (UTC)

Layer 1 (Physical) examples incorrect?[edit]

Bluetooth is given under physical layer examples in the sidebar and not in any other layers. Bluetooth goes across a lot of layers so it is inaccurate to say it's only the physical layer. Should we remove bluetooth from the examples? Alastor Moody 20:47, 7 May 2016 (UTC) — Preceding unsigned comment added by Prashantverma999 (talkcontribs)

The sentence is

The physical layer of Parallel SCSI operates in this layer, as do the physical layers of Ethernet and other local-area networks, such as Token Ring, FDDI, ITU-T, and IEEE 802.11 (Wi-Fi), as well as personal area networks such as Bluetooth and IEEE 802.15.4.

If it's parsed as

The physical layer of Parallel SCSI operates in this layer, as do the physical layers of <Ethernet and other local-area networks, such as Token Ring, FDDI, ITU-T, and IEEE 802.11 (Wi-Fi), as well as personal area networks such as Bluetooth and IEEE 802.15.4>.

i.e., with "the physical layers" referring not only to the LAN technologies but the PAN technologies, rather than as

The physical layer of Parallel SCSI operates in this layer, as do <the physical layers of Ethernet and other local-area networks, such as Token Ring, FDDI, ITU-T, and IEEE 802.11 (Wi-Fi)>, as well as personal area networks such as Bluetooth and IEEE 802.15.4.

with "the physical layers" referring only to the LAN technologies, then it's correct.
However, the sentence is complicated enough that the second parsing is plausible, and it should perhaps be rewritten to make it clear that it's referring not to Bluetooth or 802.15.4 in their entireties, but only to their physical layers, just as it's referring to the physical layers of Ethernet/Token Ring/FDDI/ rather than to those LANs in their entireties. Guy Harris (talk) 06:03, 8 May 2016 (UTC)

Layer 6: Presentation Layer - "big mapping"?[edit]

In first paragraph of Layer 6 topic, should "big mapping" be changed to "bit mapping"? Thank you. — Preceding unsigned comment added by (talk) 16:53, 20 May 2016 (UTC)

No, just to "mapping". I've done so. Guy Harris (talk) 17:24, 20 May 2016 (UTC)

Article is inconsistent on whether the presentation layer include compression and encryption[edit]

The Layer 6: Presentation Layer subtopic claims (without citation) that it is a myth that the presentation layer does compression and encryption. However, the table at the top of Description of OSI Layers mentions compression and encryption as layer 6 functions. Which is it? — Preceding unsigned comment added by Marcelocantos (talkcontribs) 23:05, 13 October 2017 (UTC)

The claim that it's a myth that compression can be done by the presentation layer is completely wrong for compression - ISO/IEC 7498-1:1994 (at least in its available-for-free X.200 version) explicitly says:
7.2.4 Functions within the Presentation Layer
The Presentation Layer performs the following functions to help accomplish the presentation-services:
a) negotiation and re-negotiation of transfer syntax;
b) representation of the abstract syntax chosen by the application-entities in the transfer syntax negotiated or renegotiated, including format and special purpose transformations (for example, data compression)
(emphasis mine). I'm not seeing anything explicit about encryption, but encryption sure sounds like a "special purpose transformation" to me. I've remove the claim about compression and, if I can find a reference, I'll rip out the claim about encryption as well. Guy Harris (talk) 00:14, 14 October 2017 (UTC)
Oh, to heck with it. I've just removed the whole "myth" claim, as, given that the "compression" part is explicitly proven wrong by the actual OSI reference model document, I see no reason to believe the "encryption" part. Guy Harris (talk) 00:18, 14 October 2017 (UTC)
X.200 (1988) also includes compression, although it also says "Recommendation X.200 and ISO 7498 [Information Processing Systems – Open Systems Interconnection – Basic Reference Model] were developed in close collaboration and are technically aligned.", so maybe they were "technically aligned" but the CCITT added compression themselves; I don't have the older version of the ISO spec handy. Perhaps the original spec didn't include it, but it was apparently added subsequently. Guy Harris (talk) 00:41, 14 October 2017 (UTC)
The confusion may be due to the fact that the OSI model, while apparently implemented, is irrelevant as far as the internet goes, apart from its use for interrogating people taking various network certification tests. My understanding (very limited) is that the presentation layer does nothing in practical systems, and that might have been behind the original research regarding "common myth". I think the OSI designers intended the presentation layer to handle things like encryption and compression, as well as EBCDIC/ASCII translation as mentioned in the article. Johnuniq (talk) 00:57, 14 October 2017 (UTC)

PHY definition[edit]

The section describing the PHY layer states: "It defines the network topology as bus, mesh, or ring being some of the most common." This is not correct. E.g. there exists both WiFi star and mesh network implementations based on the same PHY's. Another example with radio PHY's you may implement star, mesh, ad-hoc networks Cebmeisner (talk) 12:56, 15 May 2017 (UTC)

So update it to discuss radio PHYs. (The two examples you give are both radio.) Guy Harris (talk) 18:55, 15 May 2017 (UTC)