diff options
Diffstat (limited to 'web/server/h2o/libh2o/doc/benchmarks.html')
| -rw-r--r-- | web/server/h2o/libh2o/doc/benchmarks.html | 124 |
1 files changed, 124 insertions, 0 deletions
diff --git a/web/server/h2o/libh2o/doc/benchmarks.html b/web/server/h2o/libh2o/doc/benchmarks.html new file mode 100644 index 00000000..2b90b80d --- /dev/null +++ b/web/server/h2o/libh2o/doc/benchmarks.html @@ -0,0 +1,124 @@ +<!DOCTYPE html> +<html> +<head> +<meta http-equiv="Content-Type" content="text/html; charset=utf-8" /> +<meta name="viewport" content="width=device-width,initial-scale=1,user-scalable=no" /> + +<!-- oktavia --> +<link rel="stylesheet" href="assets/searchstyle.css" type="text/css" /> +<script src="search/jquery-1.9.1.min.js"></script> +<script src="search/oktavia-jquery-ui.js"></script> +<script src="search/oktavia-english-search.js"></script> +<!-- /oktavia --> + +<link rel="stylesheet" href="assets/style.css" type="text/css" /> + +<title>Benchmarks - H2O - the optimized HTTP/2 server</title> +</head> +<body> +<div id="body"> +<div id="top"> + +<h1> +<a href="index.html">H2O</a> +</h1> +<p class="description">the optimized HTTP/1.x, HTTP/2 server</p> + +<!-- oktavia --> +<form id="searchform"> +<input class="search" type="search" name="search" id="search" results="5" value="" placeholder="Search" /> +<div id="searchresult_box"> +<div id="close_search_box">×</div> +<div id="searchresult_summary"></div> +<div id="searchresult"></div> +<div id="searchresult_nav"></div> +<span class="pr">Powered by <a href="https://github.com/shibukawa/oktavia">Oktavia</a></span> +</div> +</form> +<!-- /oktavia --> + +</div> + +<table id="menu"> +<tr> +<td><a href="index.html">Top</a></td> +<td><a href="install.html">Install</a></td> +<td><a href="configure.html">Configure</a></td> +<td><a href="faq.html">FAQ</a></td> +<td><a href="http://blog.kazuhooku.com/search/label/H2O" target="_blank">Blog</a></td> +<td><a href="http://github.com/h2o/h2o/" target="_blank">Source</a></td> +</tr> +</table> + +<div id="main"> + +<h2> +Benchmarks +</h2> + + +<h3 id="download-timings">Download Timings</h3> + +<div> +<p> +Providing quick response to user is more important than anything else in web performance tuning. +According to a research conducted by Microsoft, 500msec slowdown in Bing causes their revenue go down by 1.2%<sup><a href="#note_1" id="#cite_1" title="Velocity and the Bottom Line - O'Reilly Radar">1</sup></a></sup>. +</p> +<p> +The chart below compares the first-paint times and download completion times of different web browsers / HTTP servers on a simulated network of 8Mbps bandwidth with 100ms latency, which is typical for today's mobile networks<sup><a href="#note_2" id="#cite_2" title="A fork of http2rulez.com was used as the target website; bandwidth and latency were induced to local network using qdisc, specifically by running tc qdisc replace dev eth1 root handle 1:0 tbf rate 8192kbit burst 2048 latency 100ms; sudo tc qdisc add dev eth1 parent 1:1 netem delay 100ms, and sysctl -w net.ipv4.tcp_no_metrics_save=1.">2</sup></a></sup>. +</p> +<div align="center"> +<a href="assets/8mbps100msec-nginx195-h2o150.png" target="_blank"><img src="assets/8mbps100msec-nginx195-h2o150.png" height="300"></a> +</div> +<p> +It is clear in the case of this benchmark that the visitors of the web site would be more satisfied, if H2O was used as the HTTP server. +</p> +</div> + +<h3 id="static-file">Static-File Serving</h3> + +<div> +<p> +Below chart shows the scores recorded on Amazon EC2 running two c3.8xlarge instances (server and client) on a single network placement, serving a 612-byte file<sup><a href="#note_3" id="#cite_3" title="Configuration files used: nginx.conf, h2o.conf.">3</sup></a></sup>. +For each measurement, 250 concurrent clients were used<sup><a href="#note_4" id="#cite_4" title="Wrk was used for HTTP/1 tests. h2load was used for HTTP/2.">4</sup></a></sup>. +<code>open_file_cache</code> was used for Nginx. +H2O implements a open-file-cache that gets updated immediately when the files are replaced. +</p> +<div align="center"> +<a href="assets/staticfile612-nginx1910-h2o170.png" target="_blank"><img src="assets/staticfile612-nginx1910-h2o170.png" height="300"></a> +</div> +</div> + +<h3 id="reverse-proxy">Reverse Proxy</h3> + +<div> +<p> +Presented below is an old chart showing the scores recorded on Amazon EC2 running two c3.8xlarge instances (server and client) on a single network placement<sup><a href="#note_5" id="#cite_5" title="For reverse-proxy tests, another H2O process running on the same host was used as the upstream server">5</sup></a></sup><sup><a href="#note_6" id="#cite_6" title="open-file-cache was not used in the static-file benchmark">6</sup></a></sup>. +</p> +<div align="center"> +<a href="assets/remotebench.png" target="_blank"><img src="assets/remotebench.png" width="400"></a> +</div> +</div> + + + +<div class="notes"> +<h3>Notes:</h3> +<ol> +<li id="note_1"><a href="http://radar.oreilly.com/2009/07/velocity-making-your-site-fast.html">Velocity and the Bottom Line - O'Reilly Radar</a></li> +<li id="note_2"><a href="https://github.com/kazuho/http2rulez.com">A fork of http2rulez.com</a> was used as the target website; bandwidth and latency were induced to local network using <a href="http://linux-ip.net/articles/Traffic-Control-HOWTO/components.html">qdisc</a>, specifically by running <code>tc qdisc replace dev eth1 root handle 1:0 tbf rate 8192kbit burst 2048 latency 100ms; sudo tc qdisc add dev eth1 parent 1:1 netem delay 100ms</code>, and <code>sysctl -w net.ipv4.tcp_no_metrics_save=1</code>.</li> +<li id="note_3">Configuration files used: <a href="https://gist.github.com/kazuho/def1e71281ed4ae07b95">nginx.conf</a>, <a href="https://gist.github.com/kazuho/969bb99bae31d67e01c4">h2o.conf</a>.</li> +<li id="note_4"><a href="https://github.com/wg/wrk">Wrk</a> was used for HTTP/1 tests. <a href="https://nghttp2.org/documentation/h2load-howto.html">h2load</a> was used for HTTP/2.</li> +<li id="note_5">For reverse-proxy tests, another H2O process running on the same host was used as the upstream server</li> +<li id="note_6">open-file-cache was not used in the static-file benchmark</li> +</ol> +</div> + +</div> +<div id="footer"> +<p> +Copyright © 2015 <a href="http://dena.com/intl/">DeNA Co., Ltd.</a> et al. +</p> +</div> +</body> +</html> |