blob: 7442571f58c3fe4647a6b3546ca7fc2f1450fb1e [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><a name="line.18"></a>
<span class="sourceLineNo">019</span>package org.apache.hadoop.hbase;<a name="line.19"></a>
<span class="sourceLineNo">020</span><a name="line.20"></a>
<span class="sourceLineNo">021</span>import java.io.IOException;<a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.util.concurrent.TimeUnit;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import org.apache.hadoop.conf.Configuration;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import org.apache.hadoop.fs.FSDataInputStream;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import org.apache.hadoop.fs.FileSystem;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import org.apache.hadoop.fs.Path;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.client.Result;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.client.ResultScanner;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.client.Scan;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.client.Table;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.client.TableSnapshotScanner;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.client.metrics.ScanMetrics;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.mapreduce.TableMapper;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.util.AbstractHBaseTool;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.util.CommonFSUtils;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import org.apache.hadoop.io.NullWritable;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import org.apache.hadoop.mapreduce.Job;<a name="line.42"></a>
<span class="sourceLineNo">043</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.43"></a>
<span class="sourceLineNo">044</span>import org.apache.hadoop.util.StringUtils;<a name="line.44"></a>
<span class="sourceLineNo">045</span>import org.apache.hadoop.util.ToolRunner;<a name="line.45"></a>
<span class="sourceLineNo">046</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.46"></a>
<span class="sourceLineNo">047</span><a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hbase.thirdparty.com.google.common.base.Stopwatch;<a name="line.48"></a>
<span class="sourceLineNo">049</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.49"></a>
<span class="sourceLineNo">050</span><a name="line.50"></a>
<span class="sourceLineNo">051</span>/**<a name="line.51"></a>
<span class="sourceLineNo">052</span> * A simple performance evaluation tool for single client and MR scans<a name="line.52"></a>
<span class="sourceLineNo">053</span> * and snapshot scans.<a name="line.53"></a>
<span class="sourceLineNo">054</span> */<a name="line.54"></a>
<span class="sourceLineNo">055</span>@InterfaceAudience.LimitedPrivate(HBaseInterfaceAudience.TOOLS)<a name="line.55"></a>
<span class="sourceLineNo">056</span>public class ScanPerformanceEvaluation extends AbstractHBaseTool {<a name="line.56"></a>
<span class="sourceLineNo">057</span><a name="line.57"></a>
<span class="sourceLineNo">058</span> private static final String HBASE_COUNTER_GROUP_NAME = "HBaseCounters";<a name="line.58"></a>
<span class="sourceLineNo">059</span><a name="line.59"></a>
<span class="sourceLineNo">060</span> private String type;<a name="line.60"></a>
<span class="sourceLineNo">061</span> private String file;<a name="line.61"></a>
<span class="sourceLineNo">062</span> private String tablename;<a name="line.62"></a>
<span class="sourceLineNo">063</span> private String snapshotName;<a name="line.63"></a>
<span class="sourceLineNo">064</span> private String restoreDir;<a name="line.64"></a>
<span class="sourceLineNo">065</span> private String caching;<a name="line.65"></a>
<span class="sourceLineNo">066</span><a name="line.66"></a>
<span class="sourceLineNo">067</span> @Override<a name="line.67"></a>
<span class="sourceLineNo">068</span> public void setConf(Configuration conf) {<a name="line.68"></a>
<span class="sourceLineNo">069</span> super.setConf(conf);<a name="line.69"></a>
<span class="sourceLineNo">070</span> Path rootDir;<a name="line.70"></a>
<span class="sourceLineNo">071</span> try {<a name="line.71"></a>
<span class="sourceLineNo">072</span> rootDir = CommonFSUtils.getRootDir(conf);<a name="line.72"></a>
<span class="sourceLineNo">073</span> rootDir.getFileSystem(conf);<a name="line.73"></a>
<span class="sourceLineNo">074</span> } catch (IOException ex) {<a name="line.74"></a>
<span class="sourceLineNo">075</span> throw new RuntimeException(ex);<a name="line.75"></a>
<span class="sourceLineNo">076</span> }<a name="line.76"></a>
<span class="sourceLineNo">077</span> }<a name="line.77"></a>
<span class="sourceLineNo">078</span><a name="line.78"></a>
<span class="sourceLineNo">079</span> @Override<a name="line.79"></a>
<span class="sourceLineNo">080</span> protected void addOptions() {<a name="line.80"></a>
<span class="sourceLineNo">081</span> this.addRequiredOptWithArg("t", "type", "the type of the test. One of the following: streaming|scan|snapshotscan|scanmapreduce|snapshotscanmapreduce");<a name="line.81"></a>
<span class="sourceLineNo">082</span> this.addOptWithArg("f", "file", "the filename to read from");<a name="line.82"></a>
<span class="sourceLineNo">083</span> this.addOptWithArg("tn", "table", "the tablename to read from");<a name="line.83"></a>
<span class="sourceLineNo">084</span> this.addOptWithArg("sn", "snapshot", "the snapshot name to read from");<a name="line.84"></a>
<span class="sourceLineNo">085</span> this.addOptWithArg("rs", "restoredir", "the directory to restore the snapshot");<a name="line.85"></a>
<span class="sourceLineNo">086</span> this.addOptWithArg("ch", "caching", "scanner caching value");<a name="line.86"></a>
<span class="sourceLineNo">087</span> }<a name="line.87"></a>
<span class="sourceLineNo">088</span><a name="line.88"></a>
<span class="sourceLineNo">089</span> @Override<a name="line.89"></a>
<span class="sourceLineNo">090</span> protected void processOptions(CommandLine cmd) {<a name="line.90"></a>
<span class="sourceLineNo">091</span> type = cmd.getOptionValue("type");<a name="line.91"></a>
<span class="sourceLineNo">092</span> file = cmd.getOptionValue("file");<a name="line.92"></a>
<span class="sourceLineNo">093</span> tablename = cmd.getOptionValue("table");<a name="line.93"></a>
<span class="sourceLineNo">094</span> snapshotName = cmd.getOptionValue("snapshot");<a name="line.94"></a>
<span class="sourceLineNo">095</span> restoreDir = cmd.getOptionValue("restoredir");<a name="line.95"></a>
<span class="sourceLineNo">096</span> caching = cmd.getOptionValue("caching");<a name="line.96"></a>
<span class="sourceLineNo">097</span> }<a name="line.97"></a>
<span class="sourceLineNo">098</span><a name="line.98"></a>
<span class="sourceLineNo">099</span> protected void testHdfsStreaming(Path filename) throws IOException {<a name="line.99"></a>
<span class="sourceLineNo">100</span> byte[] buf = new byte[1024];<a name="line.100"></a>
<span class="sourceLineNo">101</span> FileSystem fs = filename.getFileSystem(getConf());<a name="line.101"></a>
<span class="sourceLineNo">102</span><a name="line.102"></a>
<span class="sourceLineNo">103</span> // read the file from start to finish<a name="line.103"></a>
<span class="sourceLineNo">104</span> Stopwatch fileOpenTimer = Stopwatch.createUnstarted();<a name="line.104"></a>
<span class="sourceLineNo">105</span> Stopwatch streamTimer = Stopwatch.createUnstarted();<a name="line.105"></a>
<span class="sourceLineNo">106</span><a name="line.106"></a>
<span class="sourceLineNo">107</span> fileOpenTimer.start();<a name="line.107"></a>
<span class="sourceLineNo">108</span> FSDataInputStream in = fs.open(filename);<a name="line.108"></a>
<span class="sourceLineNo">109</span> fileOpenTimer.stop();<a name="line.109"></a>
<span class="sourceLineNo">110</span><a name="line.110"></a>
<span class="sourceLineNo">111</span> long totalBytes = 0;<a name="line.111"></a>
<span class="sourceLineNo">112</span> streamTimer.start();<a name="line.112"></a>
<span class="sourceLineNo">113</span> while (true) {<a name="line.113"></a>
<span class="sourceLineNo">114</span> int read = in.read(buf);<a name="line.114"></a>
<span class="sourceLineNo">115</span> if (read &lt; 0) {<a name="line.115"></a>
<span class="sourceLineNo">116</span> break;<a name="line.116"></a>
<span class="sourceLineNo">117</span> }<a name="line.117"></a>
<span class="sourceLineNo">118</span> totalBytes += read;<a name="line.118"></a>
<span class="sourceLineNo">119</span> }<a name="line.119"></a>
<span class="sourceLineNo">120</span> streamTimer.stop();<a name="line.120"></a>
<span class="sourceLineNo">121</span><a name="line.121"></a>
<span class="sourceLineNo">122</span> double throughput = (double)totalBytes / streamTimer.elapsed(TimeUnit.SECONDS);<a name="line.122"></a>
<span class="sourceLineNo">123</span><a name="line.123"></a>
<span class="sourceLineNo">124</span> System.out.println("HDFS streaming: ");<a name="line.124"></a>
<span class="sourceLineNo">125</span> System.out.println("total time to open: " +<a name="line.125"></a>
<span class="sourceLineNo">126</span> fileOpenTimer.elapsed(TimeUnit.MILLISECONDS) + " ms");<a name="line.126"></a>
<span class="sourceLineNo">127</span> System.out.println("total time to read: " + streamTimer.elapsed(TimeUnit.MILLISECONDS) + " ms");<a name="line.127"></a>
<span class="sourceLineNo">128</span> System.out.println("total bytes: " + totalBytes + " bytes ("<a name="line.128"></a>
<span class="sourceLineNo">129</span> + StringUtils.humanReadableInt(totalBytes) + ")");<a name="line.129"></a>
<span class="sourceLineNo">130</span> System.out.println("throghput : " + StringUtils.humanReadableInt((long)throughput) + "B/s");<a name="line.130"></a>
<span class="sourceLineNo">131</span> }<a name="line.131"></a>
<span class="sourceLineNo">132</span><a name="line.132"></a>
<span class="sourceLineNo">133</span> private Scan getScan() {<a name="line.133"></a>
<span class="sourceLineNo">134</span> Scan scan = new Scan(); // default scan settings<a name="line.134"></a>
<span class="sourceLineNo">135</span> scan.setCacheBlocks(false);<a name="line.135"></a>
<span class="sourceLineNo">136</span> scan.readVersions(1);<a name="line.136"></a>
<span class="sourceLineNo">137</span> scan.setScanMetricsEnabled(true);<a name="line.137"></a>
<span class="sourceLineNo">138</span> if (caching != null) {<a name="line.138"></a>
<span class="sourceLineNo">139</span> scan.setCaching(Integer.parseInt(caching));<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> return scan;<a name="line.142"></a>
<span class="sourceLineNo">143</span> }<a name="line.143"></a>
<span class="sourceLineNo">144</span><a name="line.144"></a>
<span class="sourceLineNo">145</span> public void testScan() throws IOException {<a name="line.145"></a>
<span class="sourceLineNo">146</span> Stopwatch tableOpenTimer = Stopwatch.createUnstarted();<a name="line.146"></a>
<span class="sourceLineNo">147</span> Stopwatch scanOpenTimer = Stopwatch.createUnstarted();<a name="line.147"></a>
<span class="sourceLineNo">148</span> Stopwatch scanTimer = Stopwatch.createUnstarted();<a name="line.148"></a>
<span class="sourceLineNo">149</span><a name="line.149"></a>
<span class="sourceLineNo">150</span> tableOpenTimer.start();<a name="line.150"></a>
<span class="sourceLineNo">151</span> Connection connection = ConnectionFactory.createConnection(getConf());<a name="line.151"></a>
<span class="sourceLineNo">152</span> Table table = connection.getTable(TableName.valueOf(tablename));<a name="line.152"></a>
<span class="sourceLineNo">153</span> tableOpenTimer.stop();<a name="line.153"></a>
<span class="sourceLineNo">154</span><a name="line.154"></a>
<span class="sourceLineNo">155</span> Scan scan = getScan();<a name="line.155"></a>
<span class="sourceLineNo">156</span> scanOpenTimer.start();<a name="line.156"></a>
<span class="sourceLineNo">157</span> ResultScanner scanner = table.getScanner(scan);<a name="line.157"></a>
<span class="sourceLineNo">158</span> scanOpenTimer.stop();<a name="line.158"></a>
<span class="sourceLineNo">159</span><a name="line.159"></a>
<span class="sourceLineNo">160</span> long numRows = 0;<a name="line.160"></a>
<span class="sourceLineNo">161</span> long numCells = 0;<a name="line.161"></a>
<span class="sourceLineNo">162</span> scanTimer.start();<a name="line.162"></a>
<span class="sourceLineNo">163</span> while (true) {<a name="line.163"></a>
<span class="sourceLineNo">164</span> Result result = scanner.next();<a name="line.164"></a>
<span class="sourceLineNo">165</span> if (result == null) {<a name="line.165"></a>
<span class="sourceLineNo">166</span> break;<a name="line.166"></a>
<span class="sourceLineNo">167</span> }<a name="line.167"></a>
<span class="sourceLineNo">168</span> numRows++;<a name="line.168"></a>
<span class="sourceLineNo">169</span><a name="line.169"></a>
<span class="sourceLineNo">170</span> numCells += result.rawCells().length;<a name="line.170"></a>
<span class="sourceLineNo">171</span> }<a name="line.171"></a>
<span class="sourceLineNo">172</span> scanTimer.stop();<a name="line.172"></a>
<span class="sourceLineNo">173</span> scanner.close();<a name="line.173"></a>
<span class="sourceLineNo">174</span> table.close();<a name="line.174"></a>
<span class="sourceLineNo">175</span> connection.close();<a name="line.175"></a>
<span class="sourceLineNo">176</span><a name="line.176"></a>
<span class="sourceLineNo">177</span> ScanMetrics metrics = scanner.getScanMetrics();<a name="line.177"></a>
<span class="sourceLineNo">178</span> long totalBytes = metrics.countOfBytesInResults.get();<a name="line.178"></a>
<span class="sourceLineNo">179</span> double throughput = (double)totalBytes / scanTimer.elapsed(TimeUnit.SECONDS);<a name="line.179"></a>
<span class="sourceLineNo">180</span> double throughputRows = (double)numRows / scanTimer.elapsed(TimeUnit.SECONDS);<a name="line.180"></a>
<span class="sourceLineNo">181</span> double throughputCells = (double)numCells / scanTimer.elapsed(TimeUnit.SECONDS);<a name="line.181"></a>
<span class="sourceLineNo">182</span><a name="line.182"></a>
<span class="sourceLineNo">183</span> System.out.println("HBase scan: ");<a name="line.183"></a>
<span class="sourceLineNo">184</span> System.out.println("total time to open table: " +<a name="line.184"></a>
<span class="sourceLineNo">185</span> tableOpenTimer.elapsed(TimeUnit.MILLISECONDS) + " ms");<a name="line.185"></a>
<span class="sourceLineNo">186</span> System.out.println("total time to open scanner: " +<a name="line.186"></a>
<span class="sourceLineNo">187</span> scanOpenTimer.elapsed(TimeUnit.MILLISECONDS) + " ms");<a name="line.187"></a>
<span class="sourceLineNo">188</span> System.out.println("total time to scan: " +<a name="line.188"></a>
<span class="sourceLineNo">189</span> scanTimer.elapsed(TimeUnit.MILLISECONDS) + " ms");<a name="line.189"></a>
<span class="sourceLineNo">190</span><a name="line.190"></a>
<span class="sourceLineNo">191</span> System.out.println("Scan metrics:\n" + metrics.getMetricsMap());<a name="line.191"></a>
<span class="sourceLineNo">192</span><a name="line.192"></a>
<span class="sourceLineNo">193</span> System.out.println("total bytes: " + totalBytes + " bytes ("<a name="line.193"></a>
<span class="sourceLineNo">194</span> + StringUtils.humanReadableInt(totalBytes) + ")");<a name="line.194"></a>
<span class="sourceLineNo">195</span> System.out.println("throughput : " + StringUtils.humanReadableInt((long)throughput) + "B/s");<a name="line.195"></a>
<span class="sourceLineNo">196</span> System.out.println("total rows : " + numRows);<a name="line.196"></a>
<span class="sourceLineNo">197</span> System.out.println("throughput : " + StringUtils.humanReadableInt((long)throughputRows) + " rows/s");<a name="line.197"></a>
<span class="sourceLineNo">198</span> System.out.println("total cells : " + numCells);<a name="line.198"></a>
<span class="sourceLineNo">199</span> System.out.println("throughput : " + StringUtils.humanReadableInt((long)throughputCells) + " cells/s");<a name="line.199"></a>
<span class="sourceLineNo">200</span> }<a name="line.200"></a>
<span class="sourceLineNo">201</span><a name="line.201"></a>
<span class="sourceLineNo">202</span><a name="line.202"></a>
<span class="sourceLineNo">203</span> public void testSnapshotScan() throws IOException {<a name="line.203"></a>
<span class="sourceLineNo">204</span> Stopwatch snapshotRestoreTimer = Stopwatch.createUnstarted();<a name="line.204"></a>
<span class="sourceLineNo">205</span> Stopwatch scanOpenTimer = Stopwatch.createUnstarted();<a name="line.205"></a>
<span class="sourceLineNo">206</span> Stopwatch scanTimer = Stopwatch.createUnstarted();<a name="line.206"></a>
<span class="sourceLineNo">207</span><a name="line.207"></a>
<span class="sourceLineNo">208</span> Path restoreDir = new Path(this.restoreDir);<a name="line.208"></a>
<span class="sourceLineNo">209</span><a name="line.209"></a>
<span class="sourceLineNo">210</span> snapshotRestoreTimer.start();<a name="line.210"></a>
<span class="sourceLineNo">211</span> restoreDir.getFileSystem(conf).delete(restoreDir, true);<a name="line.211"></a>
<span class="sourceLineNo">212</span> snapshotRestoreTimer.stop();<a name="line.212"></a>
<span class="sourceLineNo">213</span><a name="line.213"></a>
<span class="sourceLineNo">214</span> Scan scan = getScan();<a name="line.214"></a>
<span class="sourceLineNo">215</span> scanOpenTimer.start();<a name="line.215"></a>
<span class="sourceLineNo">216</span> TableSnapshotScanner scanner = new TableSnapshotScanner(conf, restoreDir, snapshotName, scan);<a name="line.216"></a>
<span class="sourceLineNo">217</span> scanOpenTimer.stop();<a name="line.217"></a>
<span class="sourceLineNo">218</span><a name="line.218"></a>
<span class="sourceLineNo">219</span> long numRows = 0;<a name="line.219"></a>
<span class="sourceLineNo">220</span> long numCells = 0;<a name="line.220"></a>
<span class="sourceLineNo">221</span> scanTimer.start();<a name="line.221"></a>
<span class="sourceLineNo">222</span> while (true) {<a name="line.222"></a>
<span class="sourceLineNo">223</span> Result result = scanner.next();<a name="line.223"></a>
<span class="sourceLineNo">224</span> if (result == null) {<a name="line.224"></a>
<span class="sourceLineNo">225</span> break;<a name="line.225"></a>
<span class="sourceLineNo">226</span> }<a name="line.226"></a>
<span class="sourceLineNo">227</span> numRows++;<a name="line.227"></a>
<span class="sourceLineNo">228</span><a name="line.228"></a>
<span class="sourceLineNo">229</span> numCells += result.rawCells().length;<a name="line.229"></a>
<span class="sourceLineNo">230</span> }<a name="line.230"></a>
<span class="sourceLineNo">231</span> scanTimer.stop();<a name="line.231"></a>
<span class="sourceLineNo">232</span> scanner.close();<a name="line.232"></a>
<span class="sourceLineNo">233</span><a name="line.233"></a>
<span class="sourceLineNo">234</span> ScanMetrics metrics = scanner.getScanMetrics();<a name="line.234"></a>
<span class="sourceLineNo">235</span> long totalBytes = metrics.countOfBytesInResults.get();<a name="line.235"></a>
<span class="sourceLineNo">236</span> double throughput = (double)totalBytes / scanTimer.elapsed(TimeUnit.SECONDS);<a name="line.236"></a>
<span class="sourceLineNo">237</span> double throughputRows = (double)numRows / scanTimer.elapsed(TimeUnit.SECONDS);<a name="line.237"></a>
<span class="sourceLineNo">238</span> double throughputCells = (double)numCells / scanTimer.elapsed(TimeUnit.SECONDS);<a name="line.238"></a>
<span class="sourceLineNo">239</span><a name="line.239"></a>
<span class="sourceLineNo">240</span> System.out.println("HBase scan snapshot: ");<a name="line.240"></a>
<span class="sourceLineNo">241</span> System.out.println("total time to restore snapshot: " +<a name="line.241"></a>
<span class="sourceLineNo">242</span> snapshotRestoreTimer.elapsed(TimeUnit.MILLISECONDS) + " ms");<a name="line.242"></a>
<span class="sourceLineNo">243</span> System.out.println("total time to open scanner: " +<a name="line.243"></a>
<span class="sourceLineNo">244</span> scanOpenTimer.elapsed(TimeUnit.MILLISECONDS) + " ms");<a name="line.244"></a>
<span class="sourceLineNo">245</span> System.out.println("total time to scan: " +<a name="line.245"></a>
<span class="sourceLineNo">246</span> scanTimer.elapsed(TimeUnit.MILLISECONDS) + " ms");<a name="line.246"></a>
<span class="sourceLineNo">247</span><a name="line.247"></a>
<span class="sourceLineNo">248</span> System.out.println("Scan metrics:\n" + metrics.getMetricsMap());<a name="line.248"></a>
<span class="sourceLineNo">249</span><a name="line.249"></a>
<span class="sourceLineNo">250</span> System.out.println("total bytes: " + totalBytes + " bytes ("<a name="line.250"></a>
<span class="sourceLineNo">251</span> + StringUtils.humanReadableInt(totalBytes) + ")");<a name="line.251"></a>
<span class="sourceLineNo">252</span> System.out.println("throughput : " + StringUtils.humanReadableInt((long)throughput) + "B/s");<a name="line.252"></a>
<span class="sourceLineNo">253</span> System.out.println("total rows : " + numRows);<a name="line.253"></a>
<span class="sourceLineNo">254</span> System.out.println("throughput : " + StringUtils.humanReadableInt((long)throughputRows) + " rows/s");<a name="line.254"></a>
<span class="sourceLineNo">255</span> System.out.println("total cells : " + numCells);<a name="line.255"></a>
<span class="sourceLineNo">256</span> System.out.println("throughput : " + StringUtils.humanReadableInt((long)throughputCells) + " cells/s");<a name="line.256"></a>
<span class="sourceLineNo">257</span><a name="line.257"></a>
<span class="sourceLineNo">258</span> }<a name="line.258"></a>
<span class="sourceLineNo">259</span><a name="line.259"></a>
<span class="sourceLineNo">260</span> public static enum ScanCounter {<a name="line.260"></a>
<span class="sourceLineNo">261</span> NUM_ROWS,<a name="line.261"></a>
<span class="sourceLineNo">262</span> NUM_CELLS,<a name="line.262"></a>
<span class="sourceLineNo">263</span> }<a name="line.263"></a>
<span class="sourceLineNo">264</span><a name="line.264"></a>
<span class="sourceLineNo">265</span> public static class MyMapper&lt;KEYOUT, VALUEOUT&gt; extends TableMapper&lt;KEYOUT, VALUEOUT&gt; {<a name="line.265"></a>
<span class="sourceLineNo">266</span> @Override<a name="line.266"></a>
<span class="sourceLineNo">267</span> protected void map(ImmutableBytesWritable key, Result value,<a name="line.267"></a>
<span class="sourceLineNo">268</span> Context context) throws IOException,<a name="line.268"></a>
<span class="sourceLineNo">269</span> InterruptedException {<a name="line.269"></a>
<span class="sourceLineNo">270</span> context.getCounter(ScanCounter.NUM_ROWS).increment(1);<a name="line.270"></a>
<span class="sourceLineNo">271</span> context.getCounter(ScanCounter.NUM_CELLS).increment(value.rawCells().length);<a name="line.271"></a>
<span class="sourceLineNo">272</span> }<a name="line.272"></a>
<span class="sourceLineNo">273</span> }<a name="line.273"></a>
<span class="sourceLineNo">274</span><a name="line.274"></a>
<span class="sourceLineNo">275</span> public void testScanMapReduce() throws IOException, InterruptedException, ClassNotFoundException {<a name="line.275"></a>
<span class="sourceLineNo">276</span> Stopwatch scanOpenTimer = Stopwatch.createUnstarted();<a name="line.276"></a>
<span class="sourceLineNo">277</span> Stopwatch scanTimer = Stopwatch.createUnstarted();<a name="line.277"></a>
<span class="sourceLineNo">278</span><a name="line.278"></a>
<span class="sourceLineNo">279</span> Scan scan = getScan();<a name="line.279"></a>
<span class="sourceLineNo">280</span><a name="line.280"></a>
<span class="sourceLineNo">281</span> String jobName = "testScanMapReduce";<a name="line.281"></a>
<span class="sourceLineNo">282</span><a name="line.282"></a>
<span class="sourceLineNo">283</span> Job job = new Job(conf);<a name="line.283"></a>
<span class="sourceLineNo">284</span> job.setJobName(jobName);<a name="line.284"></a>
<span class="sourceLineNo">285</span><a name="line.285"></a>
<span class="sourceLineNo">286</span> job.setJarByClass(getClass());<a name="line.286"></a>
<span class="sourceLineNo">287</span><a name="line.287"></a>
<span class="sourceLineNo">288</span> TableMapReduceUtil.initTableMapperJob(<a name="line.288"></a>
<span class="sourceLineNo">289</span> this.tablename,<a name="line.289"></a>
<span class="sourceLineNo">290</span> scan,<a name="line.290"></a>
<span class="sourceLineNo">291</span> MyMapper.class,<a name="line.291"></a>
<span class="sourceLineNo">292</span> NullWritable.class,<a name="line.292"></a>
<span class="sourceLineNo">293</span> NullWritable.class,<a name="line.293"></a>
<span class="sourceLineNo">294</span> job<a name="line.294"></a>
<span class="sourceLineNo">295</span> );<a name="line.295"></a>
<span class="sourceLineNo">296</span><a name="line.296"></a>
<span class="sourceLineNo">297</span> job.setNumReduceTasks(0);<a name="line.297"></a>
<span class="sourceLineNo">298</span> job.setOutputKeyClass(NullWritable.class);<a name="line.298"></a>
<span class="sourceLineNo">299</span> job.setOutputValueClass(NullWritable.class);<a name="line.299"></a>
<span class="sourceLineNo">300</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.300"></a>
<span class="sourceLineNo">301</span><a name="line.301"></a>
<span class="sourceLineNo">302</span> scanTimer.start();<a name="line.302"></a>
<span class="sourceLineNo">303</span> job.waitForCompletion(true);<a name="line.303"></a>
<span class="sourceLineNo">304</span> scanTimer.stop();<a name="line.304"></a>
<span class="sourceLineNo">305</span><a name="line.305"></a>
<span class="sourceLineNo">306</span> Counters counters = job.getCounters();<a name="line.306"></a>
<span class="sourceLineNo">307</span> long numRows = counters.findCounter(ScanCounter.NUM_ROWS).getValue();<a name="line.307"></a>
<span class="sourceLineNo">308</span> long numCells = counters.findCounter(ScanCounter.NUM_CELLS).getValue();<a name="line.308"></a>
<span class="sourceLineNo">309</span><a name="line.309"></a>
<span class="sourceLineNo">310</span> long totalBytes = counters.findCounter(HBASE_COUNTER_GROUP_NAME, "BYTES_IN_RESULTS").getValue();<a name="line.310"></a>
<span class="sourceLineNo">311</span> double throughput = (double)totalBytes / scanTimer.elapsed(TimeUnit.SECONDS);<a name="line.311"></a>
<span class="sourceLineNo">312</span> double throughputRows = (double)numRows / scanTimer.elapsed(TimeUnit.SECONDS);<a name="line.312"></a>
<span class="sourceLineNo">313</span> double throughputCells = (double)numCells / scanTimer.elapsed(TimeUnit.SECONDS);<a name="line.313"></a>
<span class="sourceLineNo">314</span><a name="line.314"></a>
<span class="sourceLineNo">315</span> System.out.println("HBase scan mapreduce: ");<a name="line.315"></a>
<span class="sourceLineNo">316</span> System.out.println("total time to open scanner: " +<a name="line.316"></a>
<span class="sourceLineNo">317</span> scanOpenTimer.elapsed(TimeUnit.MILLISECONDS) + " ms");<a name="line.317"></a>
<span class="sourceLineNo">318</span> System.out.println("total time to scan: " + scanTimer.elapsed(TimeUnit.MILLISECONDS) + " ms");<a name="line.318"></a>
<span class="sourceLineNo">319</span><a name="line.319"></a>
<span class="sourceLineNo">320</span> System.out.println("total bytes: " + totalBytes + " bytes ("<a name="line.320"></a>
<span class="sourceLineNo">321</span> + StringUtils.humanReadableInt(totalBytes) + ")");<a name="line.321"></a>
<span class="sourceLineNo">322</span> System.out.println("throughput : " + StringUtils.humanReadableInt((long)throughput) + "B/s");<a name="line.322"></a>
<span class="sourceLineNo">323</span> System.out.println("total rows : " + numRows);<a name="line.323"></a>
<span class="sourceLineNo">324</span> System.out.println("throughput : " + StringUtils.humanReadableInt((long)throughputRows) + " rows/s");<a name="line.324"></a>
<span class="sourceLineNo">325</span> System.out.println("total cells : " + numCells);<a name="line.325"></a>
<span class="sourceLineNo">326</span> System.out.println("throughput : " + StringUtils.humanReadableInt((long)throughputCells) + " cells/s");<a name="line.326"></a>
<span class="sourceLineNo">327</span> }<a name="line.327"></a>
<span class="sourceLineNo">328</span><a name="line.328"></a>
<span class="sourceLineNo">329</span> public void testSnapshotScanMapReduce() throws IOException, InterruptedException, ClassNotFoundException {<a name="line.329"></a>
<span class="sourceLineNo">330</span> Stopwatch scanOpenTimer = Stopwatch.createUnstarted();<a name="line.330"></a>
<span class="sourceLineNo">331</span> Stopwatch scanTimer = Stopwatch.createUnstarted();<a name="line.331"></a>
<span class="sourceLineNo">332</span><a name="line.332"></a>
<span class="sourceLineNo">333</span> Scan scan = getScan();<a name="line.333"></a>
<span class="sourceLineNo">334</span><a name="line.334"></a>
<span class="sourceLineNo">335</span> String jobName = "testSnapshotScanMapReduce";<a name="line.335"></a>
<span class="sourceLineNo">336</span><a name="line.336"></a>
<span class="sourceLineNo">337</span> Job job = new Job(conf);<a name="line.337"></a>
<span class="sourceLineNo">338</span> job.setJobName(jobName);<a name="line.338"></a>
<span class="sourceLineNo">339</span><a name="line.339"></a>
<span class="sourceLineNo">340</span> job.setJarByClass(getClass());<a name="line.340"></a>
<span class="sourceLineNo">341</span><a name="line.341"></a>
<span class="sourceLineNo">342</span> TableMapReduceUtil.initTableSnapshotMapperJob(<a name="line.342"></a>
<span class="sourceLineNo">343</span> this.snapshotName,<a name="line.343"></a>
<span class="sourceLineNo">344</span> scan,<a name="line.344"></a>
<span class="sourceLineNo">345</span> MyMapper.class,<a name="line.345"></a>
<span class="sourceLineNo">346</span> NullWritable.class,<a name="line.346"></a>
<span class="sourceLineNo">347</span> NullWritable.class,<a name="line.347"></a>
<span class="sourceLineNo">348</span> job,<a name="line.348"></a>
<span class="sourceLineNo">349</span> true,<a name="line.349"></a>
<span class="sourceLineNo">350</span> new Path(restoreDir)<a name="line.350"></a>
<span class="sourceLineNo">351</span> );<a name="line.351"></a>
<span class="sourceLineNo">352</span><a name="line.352"></a>
<span class="sourceLineNo">353</span> job.setNumReduceTasks(0);<a name="line.353"></a>
<span class="sourceLineNo">354</span> job.setOutputKeyClass(NullWritable.class);<a name="line.354"></a>
<span class="sourceLineNo">355</span> job.setOutputValueClass(NullWritable.class);<a name="line.355"></a>
<span class="sourceLineNo">356</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.356"></a>
<span class="sourceLineNo">357</span><a name="line.357"></a>
<span class="sourceLineNo">358</span> scanTimer.start();<a name="line.358"></a>
<span class="sourceLineNo">359</span> job.waitForCompletion(true);<a name="line.359"></a>
<span class="sourceLineNo">360</span> scanTimer.stop();<a name="line.360"></a>
<span class="sourceLineNo">361</span><a name="line.361"></a>
<span class="sourceLineNo">362</span> Counters counters = job.getCounters();<a name="line.362"></a>
<span class="sourceLineNo">363</span> long numRows = counters.findCounter(ScanCounter.NUM_ROWS).getValue();<a name="line.363"></a>
<span class="sourceLineNo">364</span> long numCells = counters.findCounter(ScanCounter.NUM_CELLS).getValue();<a name="line.364"></a>
<span class="sourceLineNo">365</span><a name="line.365"></a>
<span class="sourceLineNo">366</span> long totalBytes = counters.findCounter(HBASE_COUNTER_GROUP_NAME, "BYTES_IN_RESULTS").getValue();<a name="line.366"></a>
<span class="sourceLineNo">367</span> double throughput = (double)totalBytes / scanTimer.elapsed(TimeUnit.SECONDS);<a name="line.367"></a>
<span class="sourceLineNo">368</span> double throughputRows = (double)numRows / scanTimer.elapsed(TimeUnit.SECONDS);<a name="line.368"></a>
<span class="sourceLineNo">369</span> double throughputCells = (double)numCells / scanTimer.elapsed(TimeUnit.SECONDS);<a name="line.369"></a>
<span class="sourceLineNo">370</span><a name="line.370"></a>
<span class="sourceLineNo">371</span> System.out.println("HBase scan mapreduce: ");<a name="line.371"></a>
<span class="sourceLineNo">372</span> System.out.println("total time to open scanner: " +<a name="line.372"></a>
<span class="sourceLineNo">373</span> scanOpenTimer.elapsed(TimeUnit.MILLISECONDS) + " ms");<a name="line.373"></a>
<span class="sourceLineNo">374</span> System.out.println("total time to scan: " + scanTimer.elapsed(TimeUnit.MILLISECONDS) + " ms");<a name="line.374"></a>
<span class="sourceLineNo">375</span><a name="line.375"></a>
<span class="sourceLineNo">376</span> System.out.println("total bytes: " + totalBytes + " bytes ("<a name="line.376"></a>
<span class="sourceLineNo">377</span> + StringUtils.humanReadableInt(totalBytes) + ")");<a name="line.377"></a>
<span class="sourceLineNo">378</span> System.out.println("throughput : " + StringUtils.humanReadableInt((long)throughput) + "B/s");<a name="line.378"></a>
<span class="sourceLineNo">379</span> System.out.println("total rows : " + numRows);<a name="line.379"></a>
<span class="sourceLineNo">380</span> System.out.println("throughput : " + StringUtils.humanReadableInt((long)throughputRows) + " rows/s");<a name="line.380"></a>
<span class="sourceLineNo">381</span> System.out.println("total cells : " + numCells);<a name="line.381"></a>
<span class="sourceLineNo">382</span> System.out.println("throughput : " + StringUtils.humanReadableInt((long)throughputCells) + " cells/s");<a name="line.382"></a>
<span class="sourceLineNo">383</span> }<a name="line.383"></a>
<span class="sourceLineNo">384</span><a name="line.384"></a>
<span class="sourceLineNo">385</span> @Override<a name="line.385"></a>
<span class="sourceLineNo">386</span> protected int doWork() throws Exception {<a name="line.386"></a>
<span class="sourceLineNo">387</span> if (type.equals("streaming")) {<a name="line.387"></a>
<span class="sourceLineNo">388</span> testHdfsStreaming(new Path(file));<a name="line.388"></a>
<span class="sourceLineNo">389</span> } else if (type.equals("scan")){<a name="line.389"></a>
<span class="sourceLineNo">390</span> testScan();<a name="line.390"></a>
<span class="sourceLineNo">391</span> } else if (type.equals("snapshotscan")) {<a name="line.391"></a>
<span class="sourceLineNo">392</span> testSnapshotScan();<a name="line.392"></a>
<span class="sourceLineNo">393</span> } else if (type.equals("scanmapreduce")) {<a name="line.393"></a>
<span class="sourceLineNo">394</span> testScanMapReduce();<a name="line.394"></a>
<span class="sourceLineNo">395</span> } else if (type.equals("snapshotscanmapreduce")) {<a name="line.395"></a>
<span class="sourceLineNo">396</span> testSnapshotScanMapReduce();<a name="line.396"></a>
<span class="sourceLineNo">397</span> }<a name="line.397"></a>
<span class="sourceLineNo">398</span> return 0;<a name="line.398"></a>
<span class="sourceLineNo">399</span> }<a name="line.399"></a>
<span class="sourceLineNo">400</span><a name="line.400"></a>
<span class="sourceLineNo">401</span> public static void main (String[] args) throws Exception {<a name="line.401"></a>
<span class="sourceLineNo">402</span> int ret = ToolRunner.run(HBaseConfiguration.create(), new ScanPerformanceEvaluation(), args);<a name="line.402"></a>
<span class="sourceLineNo">403</span> System.exit(ret);<a name="line.403"></a>
<span class="sourceLineNo">404</span> }<a name="line.404"></a>
<span class="sourceLineNo">405</span>}<a name="line.405"></a>
</pre>
</div>
</body>
</html>