|This is the talk page for discussing improvements to the Server (computing) article.
This is not a forum for general discussion of the article's subject.
|Server (computing) has been listed as a level-5 vital article in Technology. If you can improve it, please do. This article has been rated as C-Class.|
|This article is of interest to the following WikiProjects:|
|Sources for development of this article may be located at|
The word "server"
I suspect that the entire article is tainted by a misunderstanding of what a server is. "Server" is simply an abbreviated form of the term "file server." The word is a noun, though it is used in noun phrases as an adjective, as nouns often are. File servers are computers; they are not "applications," nor operating systems, nor anything intangible. Unfree (talk) 04:21, 8 March 2008 (UTC)
- Getting philosophical, a server is just a general purpose machine UNTIL it is given the appropriate software that allows it to serve. In my mind, asking if a machine should be called a server, without any kind of running process, is like asking what the sound of one hand clapping is. 22.214.171.124 (talk) 14:42, 19 January 2017 (UTC)
- You are mistaken. The Oxford English Dictionary offers the following definition for a server: "In a network, any program which manages shared access to a centralized resource or service; an (often dedicated) device on which such a program is run."
- The earliest quotation they were apparently able to find is from 1972. That quotation makes it clear that the word "server" was initially used as a generality and not as a shortened version of "file server": Proc. AFIPS Conf. XL. 264/2 "This theory considers systems in which messages place demands for transmission (service) upon a single communication channel (the single server)."
- As for "file server", the earliest quotation they could find is from 1979.
- So "server" predates "file server". It is not an abbreviated form of "file server". Rather "file server" is a more specific form of the word "server" just like "web server", "mail server", "DNS server", "print server", "X server", "font server", etc. SlubGlub (talk) 18:34, 8 April 2008 (UTC)
- It is also VERY important to clarify that a "server", in fact, IS something intangible and is not a piece of hardware. The tangible hardware/computer on which a server runs is the server's "host". A server's host could be a large powerful computer or a small not-so-powerful hardware device, but the computing power or form factor of the hardware does NOT make the device itself a server. The fact that marketing departments at large computer companies have historically misused the term "server" to refer to their more powerful machines should not cause those of us in the know (i.e. computer scientists) to also misuse the term and mislead the general population (especially when they turn to this article for truth and clarity). This article can certainly explain the popular skewed use of the term "server" as being a term used to refer to more power computers and how it likely evolved, but it shouldn't "dumb down" the article simply because most people are not computer scientists.
- The supporting evidence is the same evidence if I were to use "server" to refer to a banana. A banana is not a server, and the evidence is simply the definition of "server". If the majority of the population did not understand the definition and started referring to bananas as servers they would certainly be misusing the term. This is precisely what has happened with "server" being used to refer to physical computers. The definition of server has been misunderstood thus misused. In computing, a "server" is the process running on the host computer that provides a service. This has been the definition since the beginning of ARPANET. Reference RFC-33, dated February 1970, for the definition of "host" and use of the term "serving host" to refer to the host computer running the server process. Other RFCs making reference to a "server" before the term started getting misused: RFC-123, RFC-66, RFC-80, RFC-197, and RFC-15. RFC-669 even lays out a table distinguishing the host computers from the servers they host and whether the server speaks the "Old" or "New" version of the TELNET protocol. This RFC illustrates that multiple servers can run on the same computer (host)...thus further clarification of the definition of "server" and that it refers to the process providing the service rather than to the physical computer hosting that process.
- I do not think the evidence presented is adequate for supporting your claim that the word "server" is misused to refer to a computer hosting a server. Definitions of words and terms change over time, I don't believe this is in question. Of the evidence provided, the oldest one is from 1969, most are from 1970, and latest one is from 1974. Thus, the literature only demonstrates usage of the word "server" at least 35 years ago. Modern literature (post 1990), refers to a computer hosting a server as a "server". If I am not mistaken, Computer Architecture: A Quantitative Approach uses the word server to refer to a computer, as do countless other books of equal repute. I believe that this is also the case in journals. Rilak (talk) 03:37, 12 March 2009 (UTC)
- Misuse of a word does not change its definition. If it did then by misusing "server" to refer to a banana we soon would be having a discussion about bananas being servers. My examples were specifically old to show where the term "server" stems from. This is the history and origination of the term. You can look at more recent RFCs or texts and continue to see the correct usage, but you will also see the more recent misuse. Thus, somewhere along the way someone started using (misusing) the term "server" to refer to the server's host rather than the program running on that host computer which is doing the serving. Regardless, whether the term has been misused or has gained an alternate use does not change my original point... You can not disregard history and the origins of the term "server". It is fine to mention the general population's understanding and use of the term "server" as well, but the origin, history, and first definition of "server" MUST be primary. Further, it seems the alternate use of "server" referring to a physical computer providing a service (as opposed to the process running on that computer providing the service) is still misused by hardware companies selling computers whose purpose has yet to be determined by the purchaser. What makes the computers those companies market as servers servers? Does that mean a laptop can not provide a service, because a laptop is not a server in their marketing literature? Perhaps this is yet a further evolution of the term, but if we're not careful soon we could be using an Apple to connect to a banana to serve us some fruit salad.
A few points:
- You have provided no evidence to support your claim that server is being misused. If the usage of a word to refer to a computer is such an outrage, would there not be numerous books which explicitly say that such usage is wrong? I am not seeing any quotes along the lines of, "The general population's used of the word 'server' to refer to a computer providing a service is best avoided in formal communications as it refers to a program. The correct term is 'host'."
- There are no books or quotes available stating that referring to a banana as a "server" is a misuse of the term "server", either. Does that mean that if I refer to a banana as a "server" that I am not misusing the term? Misuse of a word is simply using it in a way that is not supported by its definition. Thus, the only evidence needed is the definition itself. Unfortunately, the definition appears to be what is in question here.
- To support the definition of a "server" being a process that provides a service (as opposed to the physical computer on which that process runs), I provided many references to the etymology of the word "server". Based on the original definition and use of the word in computing, the word is being misused when referring to a physical computer. You seem to be claiming that what I refer to as a misuse is not misuse, because there is a second definition that has evolved. I agree that if there is a second definition of "server", that defines it as the physical computer doing the serving, then what I referred to as a "misuse" would not, in fact, be a misuse...however, please note that if this second definition *evolved* from the original definition (i.e. the roots of the second definition can not be traced back as far as the first definition), then at some point in time using the term in that manner would actually have been a misuse of the term. Regardless, my original comments weren't to discuss the definition of "misuse", they were to discuss the definition of "server".
- You have provided no evidence to support your claim that it is those pesky hardware vendors corrupting the technical language again with their marketing ploys. For example, when hard disk manufactures started using gigabytes to refer to one billion bytes instead of 230 bytes, everyone noticed (eventually) and much was written about their devious ways. Is there any literature that enlightens us about this outrage that is the "misuse" of the word "server" to refer to computers by harware vendors?
- I was simply providing one possible explanation behind the creation/evolution of the second definition (I don't have a detailed log of how the term "server" has gained a second usage). Again, the point was not to state how the second definition/usage of the term came to be, it was simply that it was not the primary definition.
- You ask, "What makes the computers those companies market as servers servers"? To answer, perhaps because they are configured to serve? You will find that throughout the 1990s, the same computer is marketed as a workstation and as a server, and sometimes, the only difference is the omission of multimedia hardware, the replacement of a monitor with a terminal and the bundling of server software.
- You also disregard my assertion that Hennessey, Patterson and others in the same profession use the word "server" to refer to a computer; and you continue to assert that its only the general population which misuse the word. Perhaps you are leaning towards, "Do not let the general population define words"? If it is indeed only the general population who misuse the word, then who are Hennessey, Patterson and others?
- I did no such thing. I stated that more recent RFCs, texts, and the general population do use the term differently from its primary/original definition. However, I also stated that the use of the term as it was originally defined is still used extensively. The general population will certainly evolve language over time adding definitions to existing terms, but the general population should not ignore history, etymology, or the origins of the terms they are giving new definitions to.
- I must raise the possibility that just because one particular group uses x to refer to y, it does not mean that x can also be used by another group to refer to z. This is not "wrong" or a "misuse of a word as a result of ignorance". It is quite different. Both view points can and should be covered, but none should be considered to be wrong or right. Rilak (talk) 05:09, 13 March 2009 (UTC)
- Covering both usages is precisely what I was suggesting in my original post. Unfortunately, currently (because my edits keep getting deleted) the article makes no reference to the primary definition/usage of a server being a process (computer program) that provides a service. Instead, the article is heavily weighted to state that a "server" is a piece of computer hardware. The article, in its present form, effectively ignores the etymology and primary definition of the term as originally used by Jon Postel, Vinton Cerf, Steve Crocker, and reputable computer scientists.
- Yes, all main definitions should be covered. I think the article is heavily focused on computer hardware becuase server programs are covered in more extensive detail in other articles such as Database server and Web server. The article does have numerous problems as evidenced by its tags. Rilak (talk) 04:15, 14 March 2009 (UTC)
I was just reading Comer's Internetworking with TCP/IP. In a footnote in Chapter 2, he notes: Technically, a server is a program and not a piece of hardware. However, computer users frequently (mis)apply the term to the computer responsible for running a particular server program. For example, they might say, 'That computer is our file server' when they mean 'That computer runs our file server program'. . That was written in 1993, when just about all digital electronics was already computerized. If you want to change the channel on your TV, don't say: 'Pass the remote control', but instead 'Pass the computer that runs the remote control program'. As computers become more and more specialized, it seems to me that the distinction makes less and less sense. Gah4 (talk) 16:54, 13 October 2018 (UTC)
I just noticed that I forgot to sign the one above, and it seems that sinebot didn't find it, either. I wonder, though, if a computer has server software in ROM does that make a difference? How about special purpose hardware that only has the ability to be a server for something, and can't be reprogrammed to do something else? Print servers have been dedicated ROM based for many years, likely based on simple hardware. More recently, NAS devices as file servers, likely running Linux, but otherwise not meant as general purpose processors are common. Some allow for shell login, though I suspect most people never do that, but instead use the http based configuration system. No-one would buy a print server as a replacement for a workstation. Server computers often don't have keyboard inputs and video outputs, which makes them not useful as workstations. In any case, I suspect it is close to being the WP:COMMONNAME even if CS people disagree. Gah4 (talk) 16:54, 13 October 2018 (UTC)
Rewriting the article
The article uses a very narrow notion of server, and needs to be significantly rewritten to make it complete and comprehensive. Plus, the references have to be standard technical textbooks or websites, and not generic dictionaries, books, or websites. Wiki5d (talk) 04:14, 23 November 2008 (UTC)
- I think article needs to be moved to Server (computer), which covers the hardware only. Any other use of server in the field of computing should go to Server (xyz). There is no point trying to cover multiple definitions here, when there is enough complexity in each definition to result in large articles. Rilak (talk) 13:04, 23 November 2008 (UTC)
- I agree. But I feel that this page need not moved to Server (computer), but rather appropriate portions of this page should be moved to appropriate and separate articles (some of which might not exist yet) such as Server (computer), Server operating system, Server (program), etc.; and, for every other use xyz of Server in computing, as you mentioned, there should be a separate article of the name Server (xyz). This page, Server (computing), can be made into one that summarizes all of those different uses and links to those pages as main articles. Wiki5d (talk) 06:02, 24 November 2008 (UTC)
- I Agree with Wiki5d, this seems like the best solution. There is an obvious difference between Server (computing) and Server (computer). This article should be very generic and broad - let the sub server pages deal with details of hardware/software Server permutations. I say go for it. Jwoodger (talk) 14:42, 17 July 2009 (UTC)
I agree to some points with Rilak and Wiki5d; however the "server"-article should be kept short stating that a server is a device for handling requests eg for:
- file (audio, video, text, other filetypes) sharing over internet and/or local network (FTP, SMB, AFP, UPnP AV, NFS).
- webradio broadcasting (SHOUTCAST, ...)
- webserver (for homepage, ...)
- Active directory/Bittorrent client (installing updates by a single download by server, distribution of patches over local network)
- USB or printserver
- UPS support
The server (hardware) can perhaps be instead the Mainframe article. Alternatively, the mainframe article can be cut in pieces which can be transferred to server (hardware. However, I don't like this article suggestion; as a server can be anything from a mere 100mhz webserver to a rack of quad-core 2,5ghz 64bit computers with over 8gb of RAM ....
As for the mainframe/server hardware page following is to be mentioned: 3 types of servers exist:
- servers running on Windows (supporting protocols are CIFS/SMB)
- servers running on Mac OS X (protocol is AppleTalk Filing Protocol)
- servers running on Unix/Linux (protocol is Network File System)
Commercial servers can be bought for all three types; a Comparisation of mainframes article needs to be made
- the HP ProLiant DL380 is the most popular windows server
- Apple Xserve (4core) are the most popular OSX servers
- Sun Fire X4150 and other Sun Fire servers are popular Linux/Unix Servers
It should be noted that Linux/Unix servers are more efficient/better as because of the OS, they require fewer resources for a same task. Also eg the Sun Fire servers have much better specifics as their counterparts when looked at the price 2000 vs 3000€ and 4300€, 16GB RAM vs 8GB and 2GB, comparable CPU's).
In general, servers require fast I/O, and thus use RAID5 or RAID6 on the hard disks (usually serial attached SCSI instead of SATA/eSATA)
The second best servers are OS X servers, yet these require extensive knowledge to set up and alturations (when implemented as a server with Windows clients)
Finally for the server operating system article, Windows Server 2003 R2, CentOS (?) are used
Article lead-in rewrite
I rewrote the entire lead/usage of the article to hopefully introduce the sections better. I believe this article to describe a broad concept, not a specific class of hardware, so I got rid of the hardware-specific infobox that belongs on mainframe. The last section (everyday life?) seems redundant and I think a rewrite of the three other sections may turn this article around real nicely. Please comment on lead/Usage section and decide if the article is heading in the right direction, and if so the three big remaining sections will be easy. Zab (talk) 08:49, 15 April 2009 (UTC)
- I just noticed the everyday life section and the line - "Any computer or device serving out applications or services can technically be called a server". I basically put something similar to that near the start as a more generic lead in to the whole server discussion (before defining hardware/software servers in detail). Perhaps the rest of this 'every day' section (examples, etc) could be incorporated into the usage section as a broad definition?
- Jwoodger (talk) 05:12, 17 July 2009 (UTC)
The "A server rack seen from the rear" picture
I did not take the picture and am not an expert, but in all the racks I've worked on this is not the orientation. To further support my suspicion, the viewing aspect seems off. To correct, rotate the picture 90 degrees clockwise. Mollnow (talk) 23:41, 5 January 2010 (UTC)
Merge from Server computer
Energy consumption of servers - misleading ? No question misleading and incorrect
the text states that servers need 2.5 % of the energy consumption of the US - I think this will be a misunderstanding - it should be 2.5 % of the electricity needed, not of the whole energy consumption including fuels etc. (The source is quoted correctly, the misunderstanding is there.) Plehn (talk) 16:03, 28 September 2010 (UTC)
The source is incorrect, as well. Actual 2010 consumption was 1.1-1.5% of electrical energy consumption worldwide, and 1.7-2.2% in the US per Jonathan Koomey report as reported in the NY Times, 7/31/2011. http://www.nytimes.com/2011/08/01/technology/data-centers-using-less-power-than-forecast-report-says.html?_r=0
Earlier estimates were flawed as they were for all corporate IT, including desktops and that inaccurate level has been repeated here. The ARM claim for 2020 is widely exagerated in the hopes of helping them make a commercial entry into the server market in 2014, so is thus, very biassed. It needs to be removed.
Moreover, this section lacks context in that the 6 billion cell phones in the world use more energy that all the servers in the world and pales in comparison to double-digit uses like HVAC, domestic hot water and transportation, to name a few. Finally, this section lacks perspective. It implies that this is an improper use of energy, whereas the Smart 2020 report estimates that ICT (Informantion and Communications Technology) *SAVES* 5 times its total carbon footprint in the rest of the economy (by doing things smarter). SomeGuyInOR (talk) 19:27, 18 January 2013 (UTC)
Now biased the other way... section reads as if someone has an axe to grind. Would be good to restrict to facts about consumption and leave cell phones and hot water out of it. ---Ransom (126.96.36.199 (talk) 23:21, 3 July 2013 (UTC))
Headless computer - no information about it
I got redirected to this entry from a "Headless computer" entry (found when searching the "Headless computer" term on DuckDuckGo), and now I see that there's no information here on a headless computer. — Preceding unsigned comment added by 188.8.131.52 (talk) 20:59, 12 February 2016 (UTC)
- Most servers these days are "headless computers", in that they do not have a traditional monitor, keyboard and mouse attached. They are usually accessed via ILO or remote console. - the term "headless computer" should probably be mentioned somewhere in this article if we have a redirect for it. --Versageek 21:05, 12 February 2016 (UTC)
Hardware and Remote Admin
Codename Lisa, I feel that MMC is a stretch in this instance. If we're talking about protocols, it would be RPC and SSH. If we're talking about software, it would be MMC and bash/sh/etc. Thoughts? Interference 541 (talk) 16:57, 19 June 2017 (UTC)
- Hello, Interference 541
- Thanks for adopting the Bold, refine approach. Let's me separate our points of concerns:
- I believe the article must talk about means (be it hardware or software) that are tangible to the end-user; ones that the laymen could see, even if not understand. Also, the example must either have a source or a corresponding Wikipedia article. I see your latest contribution takes care of these.
- "Browser-based out-of-band management" is one the things that short-circuits my whole thinking. If something is browser-based, then by definition, it is in-band, not out-of-band, unless "browser-based" is misnomer for "using web browser as their front-end". In case of iLO and iDRAC, it is not the browser that supports the whole affair; the foundation is a piece of hardware, right? (I admit, this topic is at the edge of my knowledge base.)
- Out-of-band management is explained a couple of paragraphs further. We can't have an in-article content fork.
- Best regards,
Codename Lisa (talk) 05:31, 20 June 2017 (UTC)
- Hey Codename Lisa,
- Thanks for replying. The in-band/out-band aspect of iDRACs and iLos falls back to the chicken/egg argument over protocols or interface. I agree that you're correct in leaving that where it is.
- With layman understanding in mind, would it be better to move the Out-of-band management reference up to the discussion of remote admin and reword things along the lines of:
- "Remote management can be conducted via various methods including Microsoft Management Console (MMC), PowerShell, and SSH. In cases where advanced administration is needed, a hardware interface can be used to perform Out-of-band management through the use of IPMI or similar technologies. Among other things, this method allows remote changes to be made that would otherwise require physical presence."
- Let me know your thoughts!
- Interference 541 (talk) 19:18, 20 June 2017 (UTC)