Misplaced Pages

HTTP pipelining: Difference between revisions

Article snapshot taken from Wikipedia with creative commons attribution-sharealike license. Give it a read and then ask your questions in the chat. We can research this topic together.
Browse history interactively← Previous editContent deleted Content addedVisualWikitext
Revision as of 22:54, 16 October 2021 editAde56facc (talk | contribs)Extended confirmed users954 editsm Summary: added bold style to words "HTTP/1.1", "superseded" and "multiplexing"← Previous edit Latest revision as of 14:59, 12 November 2024 edit undoJohn Thacker (talk | contribs)121 edits Other implementations: - Nmap supports pipelining HTTP requests 
(26 intermediate revisions by 16 users not shown)
Line 1: Line 1:
{{Short description|Computer communication technique}}
] ]
{{HTTP}} {{HTTP}}
'''HTTP pipelining''' is a feature of '''HTTP/1.1''' which allows multiple ] requests to be sent over a single ] (transmission control protocol) connection without waiting for the corresponding responses.<ref name="pipeline">{{cite web|url=http://tools.ietf.org/html/rfc7230#section-6.3.2|title=Hypertext Transfer Protocol (HTTP/1.1): Message Syntax and Routing|publisher=ietf.org|accessdate=2014-07-24}}</ref> HTTP/1.1 specification requires servers to respond to pipelined requests correctly, sending back non-pipelined but valid responses even if server does not support HTTP pipelining. Despite this requirement, many legacy HTTP/1.1 servers do not support pipelining correctly, forcing most HTTP clients to not use HTTP pipelining in practice. '''HTTP pipelining''' is a feature of '''HTTP/1.1''', which allows multiple ] requests to be sent over a single ] connection without waiting for the corresponding responses.<ref name="HTTP/1.1-pipelining">{{cite journal|url=http://tools.ietf.org/html/rfc7230#section-6.3.2|title=Hypertext Transfer Protocol (HTTP/1.1): Message Syntax and Routing: Pipelining|year=2014 |publisher=ietf.org|doi=10.17487/RFC7230 |accessdate=2014-07-24|editor-last1=Fielding |editor-last2=Reschke |editor-first1=R. |editor-first2=J. |last1=Fielding |first1=R. |last2=Reschke |first2=J. |doi-access=free }}</ref> HTTP/1.1 requires servers to respond to pipelined requests correctly, with non-pipelined but valid responses even if server does not support HTTP pipelining. Despite this requirement, many legacy HTTP/1.1 servers do not support pipelining correctly, forcing most HTTP clients to not use HTTP pipelining.


The technique was '''superseded''' by '''multiplexing''' via ],<ref name=":0">{{Cite web|url=https://developer.mozilla.org/en-US/docs/Web/HTTP/Connection_management_in_HTTP_1.x$revision/1330814|title=Revision 1330814 {{!}} Connection management in HTTP/1.x {{!}} MDN|website=MDN Web Docs|language=en-US|access-date=2018-03-19}}</ref> which is supported by most modern browsers.<ref name="browser_support">{{cite web|url=http://caniuse.com/#search=http2|title=HTTP2 browser support|accessdate=March 9, 2017}}</ref> The technique was superseded by ''']''' via ],<ref name=":0">{{Cite web|url=https://developer.mozilla.org/en-US/docs/Web/HTTP/Connection_management_in_HTTP_1.x$revision/1330814|title=Revision 1330814 {{!}} Connection management in HTTP/1.x {{!}} MDN|website=MDN Web Docs|language=en-US|access-date=2018-03-19|archive-date=2018-03-19|archive-url=https://web.archive.org/web/20180319213834/https://developer.mozilla.org/en-US/docs/Web/HTTP/Connection_management_in_HTTP_1.x$revision/1330814|url-status=dead}}</ref> which is supported by most modern ].<ref name="browser_support">{{cite web|url=http://caniuse.com/#search=http2|title=HTTP2 browser support|accessdate=March 9, 2017}}</ref>


In ], the multiplexing is accomplished through the new underlying ] transport protocol, which replaces ]. This further reduces loading time, as there is no ] anymore. In ], multiplexing is accomplished via ] which replaces ]. This further reduces loading time, as there is no ] even if some packets are lost.


==Motivation and limitations== ==Motivation and limitations==
The pipelining of requests results in a dramatic improvement<ref>{{cite web|url=http://www.w3.org/Protocols/HTTP/Performance/Pipeline.html|title=Network Performance Effects of HTTP/1.1, CSS1, and PNG|publisher=World Wide Web Consortium|accessdate=14 January 2010|date=24 June 1997|first1=Henrik Frystyk|last1=Nielsen|authorlink1=Henrik Frystyk Nielsen|first2=Jim|last2=Gettys|authorlink2=Jim Gettys|first3=Anselm|last3=Baird-Smith|first4=Eric|last4=Prud'hommeaux|first5=Håkon Wium|last5=Lie|authorlink5=Håkon Wium Lie|first6=Chris|last6=Lilley|authorlink6=Chris Lilley (computer scientist)}}</ref> in the loading times of HTML pages, especially over high ] connections such as ]s. The speedup is less apparent on broadband connections, as the limitation of HTTP 1.1 still applies: the server must send its responses in the same order that the requests were received—so the entire connection remains ]<ref name="pipeline" /> and ] can occur. The asynchronous operation of ] and ] are solutions for this.<ref name="lwnspdy">{{cite web|url=https://lwn.net/Articles/362473/|title=Reducing HTTP latency with SPDY|first=Nathan|last=Willis|date=18 November 2009|publisher=]}}</ref> Browsers ultimately did not enable pipelining by default, and by 2017 most browsers supported HTTP/2 by default which used multiplexing instead.<ref name=":0" /> The pipelining of requests results in a dramatic improvement<ref>{{cite web|url=http://www.w3.org/Protocols/HTTP/Performance/Pipeline.html|title=Network Performance Effects of HTTP/1.1, CSS1, and PNG|publisher=World Wide Web Consortium|accessdate=14 January 2010|date=24 June 1997|first1=Henrik Frystyk|last1=Nielsen|authorlink1=Henrik Frystyk Nielsen|first2=Jim|last2=Gettys|authorlink2=Jim Gettys|first3=Anselm|last3=Baird-Smith|first4=Eric|last4=Prud'hommeaux|first5=Håkon Wium|last5=Lie|authorlink5=Håkon Wium Lie|first6=Chris|last6=Lilley|authorlink6=Chris Lilley (computer scientist)}}</ref> in the loading times of HTML pages, especially over high ] connections such as ]s. The speedup is less apparent on broadband connections, as the limitation of HTTP 1.1 still applies: the server must send its responses in the same order that the requests were received—so the entire connection remains ]<ref name="HTTP/1.1-pipelining" /> and ] can occur.


The asynchronous operations of ] and ] are solution for this.<ref name="lwnspdy">{{cite web|url=https://lwn.net/Articles/362473/|title=Reducing HTTP latency with SPDY|first=Nathan|last=Willis|date=18 November 2009|publisher=]}}</ref> By 2017 most browsers supported HTTP/2 by default which uses multiplexing instead.<ref name=":0" />
Non-] requests, like those using ], should not be pipelined.<ref name="non-idempotent">{{cite web|url=http://www.w3.org/Protocols/rfc2616/rfc2616-sec8.html|title=Connections|publisher=]}}</ref> Sequences of <code>GET</code> and <code>HEAD</code> requests can always be pipelined. A sequence of other idempotent requests like <code>PUT</code> and <code>DELETE</code> can be pipelined or not depending on whether requests in the sequence depend on the effect of others.<ref name="pipeline"/>


Non-] requests such as ] should not be pipelined.<ref name="non-idempotent">{{cite web|url=http://www.w3.org/Protocols/rfc2616/rfc2616-sec8.html|title=Connections|publisher=]}}</ref> Read requests like <code>GET</code> and <code>HEAD</code> can always be pipelined. A sequence of other idempotent requests like <code>PUT</code> and <code>DELETE</code> can be pipelined or not depending on whether requests in the sequence depend on the effect of others.<ref name="HTTP/1.1-pipelining"/>
HTTP pipelining requires both the client and the server to support it. ] conforming servers are required to support pipelining. This does not mean that servers are required to pipeline responses, but that they are required not to fail if a client chooses to pipeline requests.<ref>{{cite web|url=https://www-archive.mozilla.org/projects/netlib/http/pipelining-faq.html|title=HTTP/1.1 Pipelining FAQ'}}</ref>

HTTP pipelining requires both the client and the server to support it. ] conforming servers are required to produce valid responses to pipelined requests, but may not actually process requests concurrently.<ref>{{cite web|url=https://www-archive.mozilla.org/projects/netlib/http/pipelining-faq.html|title=HTTP/1.1 Pipelining FAQ'}}</ref>
{{Clear left}} {{Clear left}}


Most pipelining problems happen in HTTP intermediate nodes (hop-by-hop), i.e. in ]s, especially in transparent proxy servers (if one of them along the HTTP chain does not handle pipelined requests properly then nothing works as it should).<ref name="make-pipelining-usable"/>
==Implementation status==
Pipelining was introduced in HTTP/1.1 and was not present in HTTP/1.0.<ref>{{Cite web |url=http://www8.org/w8-papers/5c-protocols/key/key.html |title=Archived copy |access-date=2016-04-16 |archive-url=https://web.archive.org/web/20160424204340/http://www8.org/w8-papers/5c-protocols/key/key.html |archive-date=2016-04-24 |url-status=dead }}</ref>


Using pipelining with HTTP proxy servers is usually not recommended also because the HOL blocking problem may really slow down proxy server responses (as the server responses must be in the same order of the received requests).<ref name="HTTP/1.1-pipelining"/> <ref name="MSIE-8-chat-2008"/>
It looks like that, strangely, implementing HTTP pipelining properly has never been an easy task for some developers, maybe excepted for web server developers. There have always been complains about browsers, proxies, web caches, etc. not working well when using pipelined requests / responses, up to the point that for many years (at least till 2011) software developers, engineers, etc. tried to summarize the various kind of problems they noted, to fix things and to give advices about how to deal with pipelining on the Open Web.<ref>{{Cite web|url= https://www.ietf.org/archive/id/draft-nottingham-http-pipeline-01.html|title=Making HTTP Pipelining Usable on the Open Web|date=March 14, 2011|author=Mark Nottingham|accessdate= October 16, 2021}}</ref>

'''Example''': if a client sends 4 pipelined GET requests to a proxy through a single connection and the first one is not in its cache then the proxy has to forward that request to the destination web server; if the following three requests are instead found in its cache, the proxy has to wait for the web server response, then it has to send it to the client and only then it can send the three ] responses too.

If instead a client opens 4 connections to a proxy and sends 1 GET request per connection (without using pipelining) the proxy can send the three cached responses to client in parallel before the response from server is received, decreasing the overall completion time (because requests are served in parallel with no head-of-line blocking problem).<ref name="HTTP/1.1-concurrency">{{cite journal|url=http://tools.ietf.org/html/rfc7230#section-6.4|title=Hypertext Transfer Protocol (HTTP/1.1): Message Syntax and Routing: Concurrency|year=2014 |publisher=ietf.org|doi=10.17487/RFC7230 |accessdate=2014-07-24|editor-last1=Fielding |editor-last2=Reschke |editor-first1=R. |editor-first2=J. |last1=Fielding |first1=R. |last2=Reschke |first2=J. |doi-access=free }}</ref> The same advantage exists in HTTP/2 multiplexed streams.

==Implementation status==
Pipelining was introduced in HTTP/1.1 and was not present in HTTP/1.0.<ref>{{Cite web |url=http://www8.org/w8-papers/5c-protocols/key/key.html |title=Key Differences between HTTP/1.0 and HTTP/1.1 |access-date=2016-04-16 |archive-url=https://web.archive.org/web/20160424204340/http://www8.org/w8-papers/5c-protocols/key/key.html |archive-date=2016-04-24 |url-status=dead }}</ref>


There have always been complaints about browsers, proxy servers, etc. not working well when using pipelined requests / responses, up to the point that for many years (at least till 2011) software developers, engineers, web experts, etc. tried to summarize the various kind of problems they noted, to fix things and to give advices about how to deal with pipelining on the Open Web.<ref name="make-pipelining-usable">{{Cite web|url= https://www.ietf.org/archive/id/draft-nottingham-http-pipeline-01.html|title=Making HTTP Pipelining Usable on the Open Web|date=March 14, 2011|author=Mark Nottingham|accessdate= October 16, 2021}}</ref>
===Implementation in web servers===
Implementing pipelining in ]s is a relatively simple matter of making sure that network buffers are not discarded between requests. For that reason, most modern web servers handle pipelining without any problem.


===Implementation in web browsers=== ===Implementation in web browsers===
Of all the major browsers, only ] based on ] layout engine had a fully working implementation that was enabled by default. In all other browsers HTTP pipelining was disabled or not implemented.<ref name="lwnspdy"/> Of all the major browsers, only ] had a fully working implementation that was enabled by default. In other browsers HTTP pipelining was disabled or not implemented.<ref name="lwnspdy" />
*] does not pipeline requests, due to concerns regarding buggy proxies and ].<ref>{{cite web|url=http://www.microsoft.com/windowsxp/expertzone/chats/transcripts/08_0814_ez_ie8.mspx|title=Wayback link of 'Windows Internet Explorer 8 Expert Zone Chat (August 14, 2008)'|date=August 14, 2008|publisher=]|accessdate=May 10, 2012|archive-url=https://web.archive.org/web/20101204053757/http://www.microsoft.com/windowsxp/expertzone/chats/transcripts/08_0814_ez_ie8.mspx|archive-date=December 4, 2010|url-status=dead}}</ref> *] does not support pipeline requests, due to concerns regarding buggy proxies and ].<ref name="MSIE-8-chat-2008">{{cite web|url=http://www.microsoft.com/windowsxp/expertzone/chats/transcripts/08_0814_ez_ie8.mspx|title=Wayback link of 'Windows Internet Explorer 8 Expert Zone Chat (August 14, 2008)'|date=August 14, 2008|publisher=]|accessdate=May 10, 2012|archive-url=https://web.archive.org/web/20101204053757/http://www.microsoft.com/windowsxp/expertzone/chats/transcripts/08_0814_ez_ie8.mspx|archive-date=December 4, 2010|url-status=dead}}</ref>
*] does not support pipelining.<ref>{{Cite news|url=https://blogs.msdn.microsoft.com/ie/2005/04/11/internet-explorer-and-connection-limits/|title=Internet Explorer and Connection Limits|newspaper=IEBlog|access-date=2016-11-14}}</ref> *] does not support pipelining.<ref>{{Cite news|url=https://blogs.msdn.microsoft.com/ie/2005/04/11/internet-explorer-and-connection-limits/|title=Internet Explorer and Connection Limits|newspaper=IEBlog|access-date=2016-11-14}}</ref>
*Mozilla browsers (such as ], ] and ]) support pipelining; however, it is disabled by default.<ref> ]</ref><ref>{{cite book|title=Firefox secrets|author=Cheah Chu Yeow|page=|isbn=0-9752402-4-2|url-access=registration|url=https://archive.org/details/firefoxsecrets0000chea/page/180}}</ref> Pipelining is disabled by default to avoid issues with misbehaving servers.<ref>{{cite web|url= https://bugzilla.mozilla.org/show_bug.cgi?id=264354|title=Bug 264354: Enable HTTP pipelining by default|publisher=]|accessdate=September 16, 2011}}</ref> When pipelining is enabled, Mozilla browsers use some heuristics, especially to turn pipelining off for older ] servers.<ref>{{cite web|url=https://hg.mozilla.org/releases/mozilla-2.0/file/09565753ce5f/netwerk/protocol/http/src/nsHttpConnection.cpp#l251|title=Source code – nsHttpConnection.cpp|publisher=Mozilla|work=Firefox source code|date=May 7, 2010|accessdate=December 5, 2010}}</ref> Support for H1 Pipeline was removed from Mozilla Firefox in Version 54.<ref>{{cite web|url= https://bugzilla.mozilla.org/show_bug.cgi?id=1340655|title=Bug 1340655: Remove H1 Pipeline Support|publisher=]|accessdate=March 22, 2017}}</ref> *Mozilla browsers (such as ], ] and ]) used to support pipelining; however, it was removed in Firefox 54.<ref></ref> When it was supported, pipelining was disabled by default to avoid issues with misbehaving servers.<ref>{{cite web|url= https://bugzilla.mozilla.org/show_bug.cgi?id=264354|title=Bug 264354: Enable HTTP pipelining by default|publisher=]|accessdate=September 16, 2011}}</ref> If pipelining was enabled by the user, Mozilla browsers used some heuristics, mostly to turn pipelining off for older ] servers.<ref>{{cite web|url=https://hg.mozilla.org/releases/mozilla-2.0/file/09565753ce5f/netwerk/protocol/http/src/nsHttpConnection.cpp#l251|title=Source code – nsHttpConnection.cpp|publisher=Mozilla|work=Firefox source code|date=May 7, 2010|accessdate=December 5, 2010}}</ref> The removal was eventually backported to SeaMonkey.
*] 2.0 supports pipelining, but it is disabled by default.<ref>{{Cite book|url= *] 2.0 supports pipelining, but it is disabled by default.<ref>{{Cite book|url=https://books.google.com/books?id=J1gb2eb-NuEC&dq=pipelining+%22konqueror%22&pg=PA31|title=Internet Communication: Protocols and related subjects|author=Emir Arian|date=|access-date=2021-10-16|language=en|isbn=}}</ref>
https://books.google.it/books?id=J1gb2eb-NuEC&pg=PA31&lpg=PA31&dq=pipelining+%22konqueror%22&source=bl&ots=o6qAspE_K2&sig=ACfU3U15t8hKJzvMj8Bhne3Fr6qb34NJaA&hl=it&sa=X&ved=2ahUKEwj5wfm4-s_zAhVth_0HHRfsALcQ6AF6BAgTEAM#v=onepage&q=pipelining%20%22konqueror%22&f=false|title=Internet Communication: Protocols and related subjects|author=Emir Arian|date=|access-date=2021-10-16|language=en|isbn=}}</ref>
*] previously supported pipelining, but it has been disabled due to bugs and problems with poorly behaving servers.<ref></ref> *] previously supported pipelining, but it has been disabled due to bugs and problems with poorly behaving servers.<ref></ref>
*] supports pipelining, and is enabled by default.<ref>{{Cite web|url=https://forum.palemoon.org/viewtopic.php?f=5&t=16869&p=123455|title=HTTP/1 Pipelining support has been removed in Firefox 54 - Pale Moon forum|website=forum.palemoon.org|language=en-us|access-date=2018-06-07}}</ref> *] supports pipelining, and is enabled by default.<ref>{{Cite web|url=https://forum.palemoon.org/viewtopic.php?f=5&t=16869&p=123455|title=HTTP/1 Pipelining support has been removed in Firefox 54 - Pale Moon forum|website=forum.palemoon.org|language=en-us|access-date=2018-06-07}}</ref>


===Implementation in web proxies=== ===Implementation in web proxy servers===
Most HTTP proxies do not pipeline outgoing requests.<ref>{{cite web|url= https://www.mnot.net/blog/2007/06/20/proxy_caching|title=The State of Proxy Caching|date=June 20, 2007|author=Mark Nottingham|accessdate= May 16, 2009}}</ref> Most HTTP proxies do not pipeline outgoing requests.<ref>{{cite web|url= https://www.mnot.net/blog/2007/06/20/proxy_caching|title=The State of Proxy Caching|date=June 20, 2007|author=Mark Nottingham|accessdate= May 16, 2009}}</ref>


Some HTTP proxies, including transparent HTTP proxies, may manage pipelined requests very badly (i.e. by mixing up the order of pipelined responses, etc.).<ref>{{cite web|url= https://www.mnot.net/blog/2011/07/11/what_proxies_must_do|title=What proxies must do|date=July 11, 2011|author=Mark Nottingham|accessdate= October 16, 2021}}</ref> Some HTTP proxies, including transparent HTTP proxies, may manage pipelined requests very badly (e.g. by mixing up the order of pipelined responses).<ref>{{cite web|url= https://www.mnot.net/blog/2011/07/11/what_proxies_must_do|title=What proxies must do|date=July 11, 2011|author=Mark Nottingham|accessdate= October 16, 2021}}</ref>


Some versions of the ] web proxy will pipeline up to two outgoing requests. This functionality has been disabled by default and needs to be manually enabled for "bandwidth management and access logging reasons".<ref>{{cite web|url=http://www.squid-cache.org/Doc/config/pipeline_prefetch/|title=squid : pipeline_prefetch configuration directive|date=November 9, 2009|publisher=]|accessdate=December 1, 2009}}</ref> Squid supports multiple requests from clients. Some versions of the ] web proxy will pipeline up to two outgoing requests. This functionality has been disabled by default and needs to be manually enabled for "bandwidth management and access logging reasons".<ref>{{cite web|url=http://www.squid-cache.org/Doc/config/pipeline_prefetch/|title=squid : pipeline_prefetch configuration directive|date=November 9, 2009|publisher=]|accessdate=December 1, 2009}}</ref> Squid supports multiple requests from clients.
Line 60: Line 67:
*] (a ] ] distribution system) *] (a ] ] distribution system)
*] (APT) supports pipelining.{{Citation needed|date=August 2010}} *] (APT) supports pipelining.{{Citation needed|date=August 2010}}
*] (SVN) has optional support for HTTP pipelining with the serf WebDAV access module (the default module, neon, does not have pipelining support).<ref>{{cite book | title = Version Control with Subversion|author1=C. Michael Pilato |author2=Ben Collins-Sussman |author3=Brian W. Fitzpatrick |page=238|publisher=]|year=2008|isbn=0-596-51033-0}}</ref><ref>{{cite web|url=http://www.erenkrantz.com/oscon/OSCON%202007%20Subversion%20New%20Toys.pdf|title=Subversion: Powerful New Toys|author=Justin R. Erenkrantz|year=2007}}</ref> *] (SVN) has optional support for HTTP pipelining with the serf WebDAV access module (the default module, neon, does not have pipelining support).<ref>{{cite book | title = Version Control with Subversion|author1=C. Michael Pilato |author2=Ben Collins-Sussman |author3=Brian W. Fitzpatrick |page=238|publisher=]|year=2008|isbn=978-0-596-51033-6}}</ref><ref>{{cite web|url=http://www.erenkrantz.com/oscon/OSCON%202007%20Subversion%20New%20Toys.pdf|title=Subversion: Powerful New Toys|author=Justin R. Erenkrantz|year=2007}}</ref>
*] on ] utilises pipelining on HTTP by default, and can be configured to use it on HTTPS.<ref>{{cite web|url= https://technet.microsoft.com/en-us/library/cc737394%28WS.10%29.aspx|title= HTTP/HTTPS messages|publisher=]|date=January 21, 2005}}</ref> *] on ] utilises pipelining on HTTP by default, and can be configured to use it on HTTPS.<ref>{{cite web|url= https://technet.microsoft.com/en-us/library/cc737394%28WS.10%29.aspx|title= HTTP/HTTPS messages|publisher=]|date=January 21, 2005}}</ref>
*IBM ] 3.1 supports HTTP pipelining within its client.<ref></ref> *IBM ] 3.1 supports HTTP pipelining within its client.<ref></ref>
Line 67: Line 74:


*]<ref>{{Cite web |url=http://www.hpl.hp.com/research/linux/httperf/ |title=HTTP Website |access-date=2010-10-01 |archive-url=https://web.archive.org/web/20120608143409/http://www.hpl.hp.com/research/linux/httperf/ |archive-date=2012-06-08 |url-status=dead }}</ref> *]<ref>{{Cite web |url=http://www.hpl.hp.com/research/linux/httperf/ |title=HTTP Website |access-date=2010-10-01 |archive-url=https://web.archive.org/web/20120608143409/http://www.hpl.hp.com/research/linux/httperf/ |archive-date=2012-06-08 |url-status=dead }}</ref>
*] supports pipelining requests with the Nmap Scripting Engine (NSE)<ref>{{Cite web |url=https://nmap.org/nsedoc/lib/http.html |title=Library http |access-date=2024-11-12}}</ref>


==See also== ==See also==
Line 77: Line 85:


==External links== ==External links==
*RFC 7230 {{cite web|url=http://tools.ietf.org/html/rfc7230#section-6.3.2|title=Hypertext Transfer Protocol (HTTP/1.1): Message Syntax and Routing|publisher=ietf.org|accessdate=2014-07-24}} *RFC 7230 {{cite journal|url=http://tools.ietf.org/html/rfc7230#section-6.3.2|title=Hypertext Transfer Protocol (HTTP/1.1): Message Syntax and Routing|year=2014 |publisher=ietf.org|doi=10.17487/RFC7230 |accessdate=2014-07-24|editor-last1=Fielding |editor-last2=Reschke |editor-first1=R |editor-first2=J |last1=Fielding |first1=R. |last2=Reschke |first2=J. |doi-access=free }}
* *
* *

Latest revision as of 14:59, 12 November 2024

Computer communication technique
Time diagram of non-pipelined vs. pipelined connection
HTTP
Request methods
Header fields
Response status codes
Security access control methods
Security vulnerabilities

HTTP pipelining is a feature of HTTP/1.1, which allows multiple HTTP requests to be sent over a single TCP connection without waiting for the corresponding responses. HTTP/1.1 requires servers to respond to pipelined requests correctly, with non-pipelined but valid responses even if server does not support HTTP pipelining. Despite this requirement, many legacy HTTP/1.1 servers do not support pipelining correctly, forcing most HTTP clients to not use HTTP pipelining.

The technique was superseded by multiplexing via HTTP/2, which is supported by most modern browsers.

In HTTP/3, multiplexing is accomplished via QUIC which replaces TCP. This further reduces loading time, as there is no head-of-line blocking even if some packets are lost.

Motivation and limitations

The pipelining of requests results in a dramatic improvement in the loading times of HTML pages, especially over high latency connections such as satellite Internet connections. The speedup is less apparent on broadband connections, as the limitation of HTTP 1.1 still applies: the server must send its responses in the same order that the requests were received—so the entire connection remains first-in-first-out and HOL blocking can occur.

The asynchronous operations of HTTP/2 and SPDY are solution for this. By 2017 most browsers supported HTTP/2 by default which uses multiplexing instead.

Non-idempotent requests such as POST should not be pipelined. Read requests like GET and HEAD can always be pipelined. A sequence of other idempotent requests like PUT and DELETE can be pipelined or not depending on whether requests in the sequence depend on the effect of others.

HTTP pipelining requires both the client and the server to support it. HTTP/1.1 conforming servers are required to produce valid responses to pipelined requests, but may not actually process requests concurrently.

Most pipelining problems happen in HTTP intermediate nodes (hop-by-hop), i.e. in proxy servers, especially in transparent proxy servers (if one of them along the HTTP chain does not handle pipelined requests properly then nothing works as it should).

Using pipelining with HTTP proxy servers is usually not recommended also because the HOL blocking problem may really slow down proxy server responses (as the server responses must be in the same order of the received requests).

Example: if a client sends 4 pipelined GET requests to a proxy through a single connection and the first one is not in its cache then the proxy has to forward that request to the destination web server; if the following three requests are instead found in its cache, the proxy has to wait for the web server response, then it has to send it to the client and only then it can send the three cached responses too.

If instead a client opens 4 connections to a proxy and sends 1 GET request per connection (without using pipelining) the proxy can send the three cached responses to client in parallel before the response from server is received, decreasing the overall completion time (because requests are served in parallel with no head-of-line blocking problem). The same advantage exists in HTTP/2 multiplexed streams.

Implementation status

Pipelining was introduced in HTTP/1.1 and was not present in HTTP/1.0.

There have always been complaints about browsers, proxy servers, etc. not working well when using pipelined requests / responses, up to the point that for many years (at least till 2011) software developers, engineers, web experts, etc. tried to summarize the various kind of problems they noted, to fix things and to give advices about how to deal with pipelining on the Open Web.

Implementation in web browsers

Of all the major browsers, only Opera had a fully working implementation that was enabled by default. In other browsers HTTP pipelining was disabled or not implemented.

  • Internet Explorer 8 does not support pipeline requests, due to concerns regarding buggy proxies and head-of-line blocking.
  • Internet Explorer 11 does not support pipelining.
  • Mozilla browsers (such as Mozilla Firefox, SeaMonkey and Camino) used to support pipelining; however, it was removed in Firefox 54. When it was supported, pipelining was disabled by default to avoid issues with misbehaving servers. If pipelining was enabled by the user, Mozilla browsers used some heuristics, mostly to turn pipelining off for older Microsoft IIS servers. The removal was eventually backported to SeaMonkey.
  • Konqueror 2.0 supports pipelining, but it is disabled by default.
  • Google Chrome previously supported pipelining, but it has been disabled due to bugs and problems with poorly behaving servers.
  • Pale Moon (web browser) supports pipelining, and is enabled by default.

Implementation in web proxy servers

Most HTTP proxies do not pipeline outgoing requests.

Some HTTP proxies, including transparent HTTP proxies, may manage pipelined requests very badly (e.g. by mixing up the order of pipelined responses).

Some versions of the Squid web proxy will pipeline up to two outgoing requests. This functionality has been disabled by default and needs to be manually enabled for "bandwidth management and access logging reasons". Squid supports multiple requests from clients.

The Polipo proxy pipelines outgoing requests.

Tempesta FW, an open source application delivery controller, also pipelines requests to backend servers.

Other implementations

The libwww library made by the World Wide Web Consortium (W3C), supports pipelining since version 5.1 released at 18 February 1997.

Other application development libraries that support HTTP pipelining include:

  • Perl modules providing client support for HTTP pipelining are HTTP::Async and the LWPng (libwww-perl New Generation) library.
  • The Microsoft .NET Framework 3.5 supports HTTP pipelining in the module System.Net.HttpWebRequest.
  • Qt class QNetworkRequest, introduced in 4.4.

Some other applications currently exploiting pipelining are:

  • IceBreak application server since BUILD389
  • phttpget from FreeBSD (a minimalist pipelined HTTP client)
  • libcurl previously had limited support for pipelining using the CURLMOPT_PIPELINING option, but this support was removed in version 7.65.0
  • portsnap (a FreeBSD ports tree distribution system)
  • Advanced Packaging Tool (APT) supports pipelining.
  • Subversion (SVN) has optional support for HTTP pipelining with the serf WebDAV access module (the default module, neon, does not have pipelining support).
  • Microsoft Message Queuing on Windows Server 2003 utilises pipelining on HTTP by default, and can be configured to use it on HTTPS.
  • IBM CICS 3.1 supports HTTP pipelining within its client.

Testing tools which support HTTP pipelining include:

  • httperf
  • Nmap supports pipelining requests with the Nmap Scripting Engine (NSE)

See also

References

  1. ^ Fielding, R.; Reschke, J. (2014). Fielding, R.; Reschke, J. (eds.). "Hypertext Transfer Protocol (HTTP/1.1): Message Syntax and Routing: Pipelining". ietf.org. doi:10.17487/RFC7230. Retrieved 2014-07-24. {{cite journal}}: Cite journal requires |journal= (help)
  2. ^ "Revision 1330814 | Connection management in HTTP/1.x | MDN". MDN Web Docs. Archived from the original on 2018-03-19. Retrieved 2018-03-19.
  3. "HTTP2 browser support". Retrieved March 9, 2017.
  4. Nielsen, Henrik Frystyk; Gettys, Jim; Baird-Smith, Anselm; Prud'hommeaux, Eric; Lie, Håkon Wium; Lilley, Chris (24 June 1997). "Network Performance Effects of HTTP/1.1, CSS1, and PNG". World Wide Web Consortium. Retrieved 14 January 2010.
  5. ^ Willis, Nathan (18 November 2009). "Reducing HTTP latency with SPDY". LWN.net.
  6. "Connections". w3.org.
  7. "HTTP/1.1 Pipelining FAQ'".
  8. ^ Mark Nottingham (March 14, 2011). "Making HTTP Pipelining Usable on the Open Web". Retrieved October 16, 2021.
  9. ^ "Wayback link of 'Windows Internet Explorer 8 Expert Zone Chat (August 14, 2008)'". Microsoft. August 14, 2008. Archived from the original on December 4, 2010. Retrieved May 10, 2012.
  10. Fielding, R.; Reschke, J. (2014). Fielding, R.; Reschke, J. (eds.). "Hypertext Transfer Protocol (HTTP/1.1): Message Syntax and Routing: Concurrency". ietf.org. doi:10.17487/RFC7230. Retrieved 2014-07-24. {{cite journal}}: Cite journal requires |journal= (help)
  11. "Key Differences between HTTP/1.0 and HTTP/1.1". Archived from the original on 2016-04-24. Retrieved 2016-04-16.
  12. "Internet Explorer and Connection Limits". IEBlog. Retrieved 2016-11-14.
  13. Firefox 54 Release Notes
  14. "Bug 264354: Enable HTTP pipelining by default". Mozilla. Retrieved September 16, 2011.
  15. "Source code – nsHttpConnection.cpp". Firefox source code. Mozilla. May 7, 2010. Retrieved December 5, 2010.
  16. Emir Arian. Internet Communication: Protocols and related subjects. Retrieved 2021-10-16.
  17. HTTP Pipelining - The Chromium Projects
  18. "HTTP/1 Pipelining support has been removed in Firefox 54 - Pale Moon forum". forum.palemoon.org. Retrieved 2018-06-07.
  19. Mark Nottingham (June 20, 2007). "The State of Proxy Caching". Retrieved May 16, 2009.
  20. Mark Nottingham (July 11, 2011). "What proxies must do". Retrieved October 16, 2021.
  21. "squid : pipeline_prefetch configuration directive". Squid. November 9, 2009. Retrieved December 1, 2009.
  22. "Polipo — a caching web proxy". Juliusz Chroboczek. September 18, 2009. Retrieved November 12, 2009.
  23. "Tempesta FW — a Linux Application Delivery Controller". GitHub. Retrieved March 29, 2018.
  24. "Servers: Tempesta's side - tempesta-tech/tempesta Wiki". Tempesta Technologies INC. August 1, 2017. Retrieved March 29, 2018.
  25. Kahan, José (June 7, 2002). "Change History of libwww". World Wide Web Consortium. Retrieved August 3, 2010.
  26. "Using HTTP::Async for Parallel HTTP Requests (Colin Bradford)" (PDF). Archived from the original (PDF) on 2012-03-10. Retrieved 2010-08-03.
  27. System.Net.HttpWebRequest & pipelining
  28. QNetworkRequest Class Reference Archived 2009-12-22 at the Wayback Machine, Nokia QT documentation
  29. Pipelined HTTP GET utility
  30. Curl pipelining explanation Archived 2012-06-27 at the Wayback Machine, Curl developer documentation
  31. Curl pipelining removal announcementArchived 2021-02-05 at the Wayback Machine
  32. C. Michael Pilato; Ben Collins-Sussman; Brian W. Fitzpatrick (2008). Version Control with Subversion. O'Reilly Media. p. 238. ISBN 978-0-596-51033-6.
  33. Justin R. Erenkrantz (2007). "Subversion: Powerful New Toys" (PDF).
  34. "HTTP/HTTPS messages". Microsoft TechNet. January 21, 2005.
  35. How CICS Web support handles pipelining
  36. "HTTP Website". Archived from the original on 2012-06-08. Retrieved 2010-10-01.
  37. "Library http". Retrieved 2024-11-12.

External links

Category: