<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta http-equiv="X-UA-Compatible" content="IE=edge">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>
Research | Apache Spark
</title>
<!-- Bootstrap core CSS -->
<link href="/css/cerulean.min.css" rel="stylesheet">
<link href="/css/custom.css" rel="stylesheet">
<script type="text/javascript">
<!-- Google Analytics initialization -->
var _gaq = _gaq || [];
_gaq.push(['_setAccount', 'UA-32518208-2']);
_gaq.push(['_trackPageview']);
(function() {
var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true;
ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js';
var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s);
})();
<!-- Adds slight delay to links to allow async reporting -->
function trackOutboundLink(link, category, action) {
try {
_gaq.push(['_trackEvent', category , action]);
} catch(err){}
setTimeout(function() {
document.location.href = link.href;
}, 100);
}
</script>
<!-- HTML5 shim and Respond.js IE8 support of HTML5 elements and media queries -->
<!--[if lt IE 9]>
<script src="https://oss.maxcdn.com/libs/html5shiv/3.7.0/html5shiv.js"></script>
<script src="https://oss.maxcdn.com/libs/respond.js/1.3.0/respond.min.js"></script>
<![endif]-->
</head>
<body>
<div class="container" style="max-width: 1200px;">
<div class="masthead">
<p class="lead">
<a href="/">
<img src="/images/spark-logo.png"
style="height:100px; width:auto; vertical-align: bottom; margin-top: 20px;"></a><span class="tagline">
Lightning-fast cluster computing
</span>
</p>
</div>
<nav class="navbar navbar-default" role="navigation">
<!-- Brand and toggle get grouped for better mobile display -->
<div class="navbar-header">
<button type="button" class="navbar-toggle" data-toggle="collapse"
data-target="#navbar-collapse-1">
<span class="sr-only">Toggle navigation</span>
<span class="icon-bar"></span>
<span class="icon-bar"></span>
<span class="icon-bar"></span>
</button>
</div>
<!-- Collect the nav links, forms, and other content for toggling -->
<div class="collapse navbar-collapse" id="navbar-collapse-1">
<ul class="nav navbar-nav">
<li><a href="/downloads.html">Download</a></li>
<li class="dropdown">
<a href="#" class="dropdown-toggle" data-toggle="dropdown">
Related Projects <b class="caret"></b>
</a>
<ul class="dropdown-menu">
<li><a href="http://shark.cs.berkeley.edu">Shark (SQL)</a></li>
<li><a href="/streaming/">Spark Streaming</a></li>
<li><a href="/mllib/">MLlib (machine learning)</a></li>
<li><a href="http://amplab.github.io/graphx/">GraphX (graph)</a></li>
</ul>
</li>
<li class="dropdown">
<a href="#" class="dropdown-toggle" data-toggle="dropdown">
Documentation <b class="caret"></b>
</a>
<ul class="dropdown-menu">
<li><a href="/documentation.html">Overview</a></li>
<li><a href="/docs/latest/">Latest Release</a></li>
<li><a href="/examples.html">Examples</a></li>
</ul>
</li>
<li class="dropdown">
<a href="#" class="dropdown-toggle" data-toggle="dropdown">
Community <b class="caret"></b>
</a>
<ul class="dropdown-menu">
<li><a href="/community.html">Mailing Lists</a></li>
<li><a href="/community.html#events">Events and Meetups</a></li>
<li><a href="/community.html#history">Project History</a></li>
<li><a href="https://cwiki.apache.org/confluence/display/SPARK/Powered+By+Spark">Powered By</a></li>
</ul>
</li>
<li><a href="/faq.html">FAQ</a></li>
</ul>
</div>
<!-- /.navbar-collapse -->
</nav>
<div class="row">
<div class="col-md-3 col-md-push-9">
<div class="news" style="margin-bottom: 20px;">
<h5>Latest News</h5>
<ul class="list-unstyled">
<li><a href="/news/spark-0-8-1-released.html">Spark 0.8.1 released</a>
<span class="small">(Dec 19, 2013)</span></li>
<li><a href="/news/spark-summit-2013-is-a-wrap.html">Spark Summit 2013 is a Wrap</a>
<span class="small">(Dec 15, 2013)</span></li>
<li><a href="/news/announcing-the-first-spark-summit.html">Announcing the first Spark Summit: December 2, 2013</a>
<span class="small">(Oct 08, 2013)</span></li>
<li><a href="/news/spark-0-8-0-released.html">Spark 0.8.0 released</a>
<span class="small">(Sep 25, 2013)</span></li>
</ul>
<p class="small" style="text-align: right;"><a href="/news/index.html">Archive</a></p>
</div>
<div class="hidden-xs hidden-sm">
<a href="/downloads.html" class="btn btn-success btn-lg btn-block" style="margin-bottom: 30px;">
Download Spark
</a>
<p style="font-size: 16px; font-weight: 500; color: #555;">
Related Projects:
</p>
<ul class="list-narrow">
<li><a href="http://shark.cs.berkeley.edu">Shark (SQL)</a></li>
<li><a href="/streaming/">Spark Streaming</a></li>
<li><a href="/mllib/">MLlib (machine learning)</a></li>
<li><a href="http://amplab.github.io/graphx/">GraphX (graph)</a></li>
</ul>
</div>
</div>
<div class="col-md-9 col-md-pull-3">
<h2>Spark Research</h2>
<p>
Apache Spark started as a research project at UC Berkeley in the <a href="https://amplab.cs.berkeley.edu">AMPLab</a>, which focuses on big data analytics.
</p>
<p class="noskip">
Our goal was to design a programming model that supports a much wider class of applications than MapReduce, while maintaining its automatic fault tolerance. In particular, MapReduce is inefficient for <em>multi-pass</em> applications that require low-latency data sharing across multiple parallel operations. These applications are quite common in analytics, and include:
</p>
<ul>
<li><em>Iterative algorithms</em>, including many machine learning algorithms and graph algorithms like PageRank.</li>
<li><em>Interactive data mining</em>, where a user would like to load data into RAM across a cluster and query it repeatedly.</li>
<li><em>Streaming applications</em> that maintain aggregate state over time.</li>
</ul>
<p>
Traditional MapReduce and DAG engines are suboptimal for these applications because they are based on acyclic data flow: an application has to run as a series of distinct jobs, each of which reads data from stable storage (e.g. a distributed file system) and writes it back to stable storage. They incur significant cost loading the data on each step and writing it back to replicated storage.
</p>
<p>
Spark offers an abstraction called <a href="http://www.cs.berkeley.edu/~matei/papers/2012/nsdi_spark.pdf"><em>resilient distributed datasets (RDDs)</em></a> to support these applications efficiently. RDDs can be stored in memory between queries <em>without</em> requiring replication. Instead, they rebuild lost data on failure using <em>lineage</em>: each RDD remembers how it was built from other datasets (by transformations like <code>map</code>, <code>join</code> or <code>groupBy</code>) to rebuild itself. RDDs allow Spark to outperform existing models by up to 100x in multi-pass analytics. We showed that RDDs can support a wide variety of iterative algorithms, as well as interactive data mining and a highly efficient SQL engine (<a href="http://shark.cs.berkeley.edu">Shark</a>).
</p>
<p class="noskip">You can find more about the research behind Spark in the following papers:</p>
<ul>
<li>
<a href="http://www.cs.berkeley.edu/~matei/papers/2013/sosp_spark_streaming.pdf">Discretized Streams: Fault-Tolerant Streaming Computation at Scale</a>. Matei Zaharia, Tathagata Das, Haoyuan Li, Timothy Hunter, Scott Shenker, Ion Stoica. <em>SOSP 2013</em>. November 2013.
</li>
<li>
<a href="http://www.cs.berkeley.edu/~matei/papers/2013/sigmod_shark.pdf">Shark: SQL and Rich Analytics at Scale</a>. Reynold Xin, Joshua Rosen, Matei Zaharia, Michael J. Franklin, Scott Shenker, Ion Stoica. <em>SIGMOD 2013</em>. June 2013.
</li>
<li>
<a href="http://www.cs.berkeley.edu/~matei/papers/2012/hotcloud_spark_streaming.pdf">Discretized Streams: An Efficient and Fault-Tolerant Model for Stream Processing on Large Clusters</a>. Matei Zaharia, Tathagata Das, Haoyuan Li, Scott Shenker, Ion Stoica. <em>HotCloud 2012</em>. June 2012.
</li>
<li>
<a href="http://www.cs.berkeley.edu/~matei/papers/2012/sigmod_shark_demo.pdf">Shark: Fast Data Analysis Using Coarse-grained Distributed Memory</a> (demo). Cliff Engle, Antonio Lupher, Reynold Xin, Matei Zaharia, Haoyuan Li, Scott Shenker, Ion Stoica. <em>SIGMOD 2012</em>. May 2012. <b>Best Demo Award</b>.
</li>
<li>
<a href="http://www.cs.berkeley.edu/~matei/papers/2012/nsdi_spark.pdf">Resilient Distributed Datasets: A Fault-Tolerant Abstraction for In-Memory Cluster Computing</a>. Matei Zaharia, Mosharaf Chowdhury, Tathagata Das, Ankur Dave, Justin Ma, Murphy McCauley, Michael J. Franklin, Scott Shenker, Ion Stoica. <em>NSDI 2012</em>. April 2012. <b>Best Paper Award</b> and <b>Honorable Mention for Community Award</b>.
</li>
<li>
<a href="http://www.cs.berkeley.edu/~matei/papers/2010/hotcloud_spark.pdf">Spark: Cluster Computing with Working Sets</a>. Matei Zaharia, Mosharaf Chowdhury, Michael J. Franklin, Scott Shenker, Ion Stoica. <em>HotCloud 2010</em>. June 2010.
</li>
</ul>
</div>
</div>
<footer class="small">
<hr>
Apache Spark is an effort undergoing incubation at The Apache Software Foundation.
<a href="http://incubator.apache.org/" style="border: none;">
<img style="vertical-align: middle; float: right; margin-bottom: 15px;"
src="/images/incubator-logo.png" alt="Apache Incubator" title="Apache Incubator" />
</a>
</footer>
</div>
<script src="https://code.jquery.com/jquery.js"></script>
<script src="//netdna.bootstrapcdn.com/bootstrap/3.0.3/js/bootstrap.min.js"></script>
<script src="/js/lang-tabs.js"></script>
</body>
</html>