Google Git
Sign in
apache / hbase-site / 943f542910cfac61bd3b7900ce5c091baa4a3064 / . / testdevapidocs / src-html / org / apache / hadoop / hbase / regionserver / EncodedSeekPerformanceTest.html
blob: ed3be4ed21e7990e92514165301b4e5228b1692a [file] [log] [blame]
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html lang="en">
<head>
<title>Source code</title>
<link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style">
</head>
<body>
<div class="sourceContainer">
<pre><span class="sourceLineNo">001</span>/*<a name="line.1"></a>
<span class="sourceLineNo">002</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.2"></a>
<span class="sourceLineNo">003</span> * or more contributor license agreements. See the NOTICE file<a name="line.3"></a>
<span class="sourceLineNo">004</span> * distributed with this work for additional information<a name="line.4"></a>
<span class="sourceLineNo">005</span> * regarding copyright ownership. The ASF licenses this file<a name="line.5"></a>
<span class="sourceLineNo">006</span> * to you under the Apache License, Version 2.0 (the<a name="line.6"></a>
<span class="sourceLineNo">007</span> * "License"); you may not use this file except in compliance<a name="line.7"></a>
<span class="sourceLineNo">008</span> * with the License. You may obtain a copy of the License at<a name="line.8"></a>
<span class="sourceLineNo">009</span> *<a name="line.9"></a>
<span class="sourceLineNo">010</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.10"></a>
<span class="sourceLineNo">011</span> *<a name="line.11"></a>
<span class="sourceLineNo">012</span> * Unless required by applicable law or agreed to in writing, software<a name="line.12"></a>
<span class="sourceLineNo">013</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.13"></a>
<span class="sourceLineNo">014</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.14"></a>
<span class="sourceLineNo">015</span> * See the License for the specific language governing permissions and<a name="line.15"></a>
<span class="sourceLineNo">016</span> * limitations under the License.<a name="line.16"></a>
<span class="sourceLineNo">017</span> */<a name="line.17"></a>
<span class="sourceLineNo">018</span>package org.apache.hadoop.hbase.regionserver;<a name="line.18"></a>
<span class="sourceLineNo">019</span><a name="line.19"></a>
<span class="sourceLineNo">020</span>import java.io.IOException;<a name="line.20"></a>
<span class="sourceLineNo">021</span>import java.util.ArrayList;<a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.util.List;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import java.util.Random;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import org.apache.hadoop.conf.Configuration;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import org.apache.hadoop.fs.Path;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.Cell;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.HBaseTestingUtil;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.HConstants;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.KeyValue;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.KeyValueUtil;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.io.hfile.CacheConfig;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.io.hfile.LruBlockCache;<a name="line.33"></a>
<span class="sourceLineNo">034</span><a name="line.34"></a>
<span class="sourceLineNo">035</span>/**<a name="line.35"></a>
<span class="sourceLineNo">036</span> * Test seek performance for encoded data blocks. Read an HFile and do several random seeks.<a name="line.36"></a>
<span class="sourceLineNo">037</span> */<a name="line.37"></a>
<span class="sourceLineNo">038</span>public class EncodedSeekPerformanceTest {<a name="line.38"></a>
<span class="sourceLineNo">039</span> private static final double NANOSEC_IN_SEC = 1000.0 * 1000.0 * 1000.0;<a name="line.39"></a>
<span class="sourceLineNo">040</span> private static final double BYTES_IN_MEGABYTES = 1024.0 * 1024.0;<a name="line.40"></a>
<span class="sourceLineNo">041</span> /** Default number of seeks which will be used in benchmark. */<a name="line.41"></a>
<span class="sourceLineNo">042</span> public static int DEFAULT_NUMBER_OF_SEEKS = 10000;<a name="line.42"></a>
<span class="sourceLineNo">043</span><a name="line.43"></a>
<span class="sourceLineNo">044</span> private final HBaseTestingUtil testingUtility = new HBaseTestingUtil();<a name="line.44"></a>
<span class="sourceLineNo">045</span> private Configuration configuration = testingUtility.getConfiguration();<a name="line.45"></a>
<span class="sourceLineNo">046</span> private CacheConfig cacheConf = new CacheConfig(configuration);<a name="line.46"></a>
<span class="sourceLineNo">047</span> private Random randomizer;<a name="line.47"></a>
<span class="sourceLineNo">048</span> private int numberOfSeeks;<a name="line.48"></a>
<span class="sourceLineNo">049</span><a name="line.49"></a>
<span class="sourceLineNo">050</span> /** Use this benchmark with default options */<a name="line.50"></a>
<span class="sourceLineNo">051</span> public EncodedSeekPerformanceTest() {<a name="line.51"></a>
<span class="sourceLineNo">052</span> configuration.setFloat(HConstants.HFILE_BLOCK_CACHE_SIZE_KEY, 0.5f);<a name="line.52"></a>
<span class="sourceLineNo">053</span> randomizer = new Random(42L);<a name="line.53"></a>
<span class="sourceLineNo">054</span> numberOfSeeks = DEFAULT_NUMBER_OF_SEEKS;<a name="line.54"></a>
<span class="sourceLineNo">055</span> }<a name="line.55"></a>
<span class="sourceLineNo">056</span><a name="line.56"></a>
<span class="sourceLineNo">057</span> private List&lt;Cell&gt; prepareListOfTestSeeks(Path path) throws IOException {<a name="line.57"></a>
<span class="sourceLineNo">058</span> List&lt;Cell&gt; allKeyValues = new ArrayList&lt;&gt;();<a name="line.58"></a>
<span class="sourceLineNo">059</span><a name="line.59"></a>
<span class="sourceLineNo">060</span> // read all of the key values<a name="line.60"></a>
<span class="sourceLineNo">061</span> HStoreFile storeFile = new HStoreFile(testingUtility.getTestFileSystem(), path, configuration,<a name="line.61"></a>
<span class="sourceLineNo">062</span> cacheConf, BloomType.NONE, true);<a name="line.62"></a>
<span class="sourceLineNo">063</span> storeFile.initReader();<a name="line.63"></a>
<span class="sourceLineNo">064</span> StoreFileReader reader = storeFile.getReader();<a name="line.64"></a>
<span class="sourceLineNo">065</span> StoreFileScanner scanner = reader.getStoreFileScanner(true, false, false, 0, 0, false);<a name="line.65"></a>
<span class="sourceLineNo">066</span> Cell current;<a name="line.66"></a>
<span class="sourceLineNo">067</span><a name="line.67"></a>
<span class="sourceLineNo">068</span> scanner.seek(KeyValue.LOWESTKEY);<a name="line.68"></a>
<span class="sourceLineNo">069</span> while (null != (current = scanner.next())) {<a name="line.69"></a>
<span class="sourceLineNo">070</span> allKeyValues.add(current);<a name="line.70"></a>
<span class="sourceLineNo">071</span> }<a name="line.71"></a>
<span class="sourceLineNo">072</span><a name="line.72"></a>
<span class="sourceLineNo">073</span> storeFile.closeStoreFile(cacheConf.shouldEvictOnClose());<a name="line.73"></a>
<span class="sourceLineNo">074</span><a name="line.74"></a>
<span class="sourceLineNo">075</span> // pick seeks by random<a name="line.75"></a>
<span class="sourceLineNo">076</span> List&lt;Cell&gt; seeks = new ArrayList&lt;&gt;();<a name="line.76"></a>
<span class="sourceLineNo">077</span> for (int i = 0; i &lt; numberOfSeeks; ++i) {<a name="line.77"></a>
<span class="sourceLineNo">078</span> Cell keyValue = allKeyValues.get(randomizer.nextInt(allKeyValues.size()));<a name="line.78"></a>
<span class="sourceLineNo">079</span> seeks.add(keyValue);<a name="line.79"></a>
<span class="sourceLineNo">080</span> }<a name="line.80"></a>
<span class="sourceLineNo">081</span><a name="line.81"></a>
<span class="sourceLineNo">082</span> clearBlockCache();<a name="line.82"></a>
<span class="sourceLineNo">083</span><a name="line.83"></a>
<span class="sourceLineNo">084</span> return seeks;<a name="line.84"></a>
<span class="sourceLineNo">085</span> }<a name="line.85"></a>
<span class="sourceLineNo">086</span><a name="line.86"></a>
<span class="sourceLineNo">087</span> private void runTest(Path path, DataBlockEncoding blockEncoding, List&lt;Cell&gt; seeks)<a name="line.87"></a>
<span class="sourceLineNo">088</span> throws IOException {<a name="line.88"></a>
<span class="sourceLineNo">089</span> // read all of the key values<a name="line.89"></a>
<span class="sourceLineNo">090</span> HStoreFile storeFile = new HStoreFile(testingUtility.getTestFileSystem(), path, configuration,<a name="line.90"></a>
<span class="sourceLineNo">091</span> cacheConf, BloomType.NONE, true);<a name="line.91"></a>
<span class="sourceLineNo">092</span> storeFile.initReader();<a name="line.92"></a>
<span class="sourceLineNo">093</span> long totalSize = 0;<a name="line.93"></a>
<span class="sourceLineNo">094</span><a name="line.94"></a>
<span class="sourceLineNo">095</span> StoreFileReader reader = storeFile.getReader();<a name="line.95"></a>
<span class="sourceLineNo">096</span> StoreFileScanner scanner = reader.getStoreFileScanner(true, false, false, 0, 0, false);<a name="line.96"></a>
<span class="sourceLineNo">097</span><a name="line.97"></a>
<span class="sourceLineNo">098</span> long startReadingTime = System.nanoTime();<a name="line.98"></a>
<span class="sourceLineNo">099</span> Cell current;<a name="line.99"></a>
<span class="sourceLineNo">100</span> scanner.seek(KeyValue.LOWESTKEY);<a name="line.100"></a>
<span class="sourceLineNo">101</span> while (null != (current = scanner.next())) { // just iterate it!<a name="line.101"></a>
<span class="sourceLineNo">102</span> if (KeyValueUtil.ensureKeyValue(current).getLength() &lt; 0) {<a name="line.102"></a>
<span class="sourceLineNo">103</span> throw new IOException("Negative KV size: " + current);<a name="line.103"></a>
<span class="sourceLineNo">104</span> }<a name="line.104"></a>
<span class="sourceLineNo">105</span> totalSize += KeyValueUtil.ensureKeyValue(current).getLength();<a name="line.105"></a>
<span class="sourceLineNo">106</span> }<a name="line.106"></a>
<span class="sourceLineNo">107</span> long finishReadingTime = System.nanoTime();<a name="line.107"></a>
<span class="sourceLineNo">108</span><a name="line.108"></a>
<span class="sourceLineNo">109</span> // do seeks<a name="line.109"></a>
<span class="sourceLineNo">110</span> long startSeeksTime = System.nanoTime();<a name="line.110"></a>
<span class="sourceLineNo">111</span> for (Cell keyValue : seeks) {<a name="line.111"></a>
<span class="sourceLineNo">112</span> scanner.seek(keyValue);<a name="line.112"></a>
<span class="sourceLineNo">113</span> Cell toVerify = scanner.next();<a name="line.113"></a>
<span class="sourceLineNo">114</span> if (!keyValue.equals(toVerify)) {<a name="line.114"></a>
<span class="sourceLineNo">115</span> System.out<a name="line.115"></a>
<span class="sourceLineNo">116</span> .println(String.format("KeyValue doesn't match:\n" + "Orig key: %s\n" + "Ret key: %s",<a name="line.116"></a>
<span class="sourceLineNo">117</span> KeyValueUtil.ensureKeyValue(keyValue).getKeyString(),<a name="line.117"></a>
<span class="sourceLineNo">118</span> KeyValueUtil.ensureKeyValue(toVerify).getKeyString()));<a name="line.118"></a>
<span class="sourceLineNo">119</span> break;<a name="line.119"></a>
<span class="sourceLineNo">120</span> }<a name="line.120"></a>
<span class="sourceLineNo">121</span> }<a name="line.121"></a>
<span class="sourceLineNo">122</span> long finishSeeksTime = System.nanoTime();<a name="line.122"></a>
<span class="sourceLineNo">123</span> if (finishSeeksTime &lt; startSeeksTime) {<a name="line.123"></a>
<span class="sourceLineNo">124</span> throw new AssertionError(<a name="line.124"></a>
<span class="sourceLineNo">125</span> "Finish time " + finishSeeksTime + " is earlier than start time " + startSeeksTime);<a name="line.125"></a>
<span class="sourceLineNo">126</span> }<a name="line.126"></a>
<span class="sourceLineNo">127</span><a name="line.127"></a>
<span class="sourceLineNo">128</span> // write some stats<a name="line.128"></a>
<span class="sourceLineNo">129</span> double readInMbPerSec =<a name="line.129"></a>
<span class="sourceLineNo">130</span> (totalSize * NANOSEC_IN_SEC) / (BYTES_IN_MEGABYTES * (finishReadingTime - startReadingTime));<a name="line.130"></a>
<span class="sourceLineNo">131</span> double seeksPerSec = (seeks.size() * NANOSEC_IN_SEC) / (finishSeeksTime - startSeeksTime);<a name="line.131"></a>
<span class="sourceLineNo">132</span><a name="line.132"></a>
<span class="sourceLineNo">133</span> storeFile.closeStoreFile(cacheConf.shouldEvictOnClose());<a name="line.133"></a>
<span class="sourceLineNo">134</span> clearBlockCache();<a name="line.134"></a>
<span class="sourceLineNo">135</span><a name="line.135"></a>
<span class="sourceLineNo">136</span> System.out.println(blockEncoding);<a name="line.136"></a>
<span class="sourceLineNo">137</span> System.out.printf(" Read speed: %8.2f (MB/s)\n", readInMbPerSec);<a name="line.137"></a>
<span class="sourceLineNo">138</span> System.out.printf(" Seeks per second: %8.2f (#/s)\n", seeksPerSec);<a name="line.138"></a>
<span class="sourceLineNo">139</span> System.out.printf(" Total KV size: %d\n", totalSize);<a name="line.139"></a>
<span class="sourceLineNo">140</span> }<a name="line.140"></a>
<span class="sourceLineNo">141</span><a name="line.141"></a>
<span class="sourceLineNo">142</span> /**<a name="line.142"></a>
<span class="sourceLineNo">143</span> * @param path Path to the HFile which will be used.<a name="line.143"></a>
<span class="sourceLineNo">144</span> * @param encodings the data block encoding algorithms to use<a name="line.144"></a>
<span class="sourceLineNo">145</span> * @throws IOException if there is a bug while reading from disk<a name="line.145"></a>
<span class="sourceLineNo">146</span> */<a name="line.146"></a>
<span class="sourceLineNo">147</span> public void runTests(Path path, DataBlockEncoding[] encodings) throws IOException {<a name="line.147"></a>
<span class="sourceLineNo">148</span> List&lt;Cell&gt; seeks = prepareListOfTestSeeks(path);<a name="line.148"></a>
<span class="sourceLineNo">149</span><a name="line.149"></a>
<span class="sourceLineNo">150</span> for (DataBlockEncoding blockEncoding : encodings) {<a name="line.150"></a>
<span class="sourceLineNo">151</span> runTest(path, blockEncoding, seeks);<a name="line.151"></a>
<span class="sourceLineNo">152</span> }<a name="line.152"></a>
<span class="sourceLineNo">153</span> }<a name="line.153"></a>
<span class="sourceLineNo">154</span><a name="line.154"></a>
<span class="sourceLineNo">155</span> /**<a name="line.155"></a>
<span class="sourceLineNo">156</span> * Command line interface:<a name="line.156"></a>
<span class="sourceLineNo">157</span> * @param args Takes one argument - file size.<a name="line.157"></a>
<span class="sourceLineNo">158</span> * @throws IOException if there is a bug while reading from disk<a name="line.158"></a>
<span class="sourceLineNo">159</span> */<a name="line.159"></a>
<span class="sourceLineNo">160</span> public static void main(final String[] args) throws IOException {<a name="line.160"></a>
<span class="sourceLineNo">161</span> if (args.length &lt; 1) {<a name="line.161"></a>
<span class="sourceLineNo">162</span> printUsage();<a name="line.162"></a>
<span class="sourceLineNo">163</span> System.exit(-1);<a name="line.163"></a>
<span class="sourceLineNo">164</span> }<a name="line.164"></a>
<span class="sourceLineNo">165</span><a name="line.165"></a>
<span class="sourceLineNo">166</span> Path path = new Path(args[0]);<a name="line.166"></a>
<span class="sourceLineNo">167</span><a name="line.167"></a>
<span class="sourceLineNo">168</span> // TODO, this test doesn't work as expected any more. Need to fix.<a name="line.168"></a>
<span class="sourceLineNo">169</span> EncodedSeekPerformanceTest utility = new EncodedSeekPerformanceTest();<a name="line.169"></a>
<span class="sourceLineNo">170</span> utility.runTests(path, DataBlockEncoding.values());<a name="line.170"></a>
<span class="sourceLineNo">171</span><a name="line.171"></a>
<span class="sourceLineNo">172</span> System.exit(0);<a name="line.172"></a>
<span class="sourceLineNo">173</span> }<a name="line.173"></a>
<span class="sourceLineNo">174</span><a name="line.174"></a>
<span class="sourceLineNo">175</span> private static void printUsage() {<a name="line.175"></a>
<span class="sourceLineNo">176</span> System.out.println("Usage: one argument, name of the HFile");<a name="line.176"></a>
<span class="sourceLineNo">177</span> }<a name="line.177"></a>
<span class="sourceLineNo">178</span><a name="line.178"></a>
<span class="sourceLineNo">179</span> private void clearBlockCache() {<a name="line.179"></a>
<span class="sourceLineNo">180</span> ((LruBlockCache) cacheConf.getBlockCache().get()).clearCache();<a name="line.180"></a>
<span class="sourceLineNo">181</span> }<a name="line.181"></a>
<span class="sourceLineNo">182</span>}<a name="line.182"></a>
</pre>
</div>
</body>
</html>