脱离Hadoop的环境下使用Lzo

动机:希望通过Java API将日志行压缩到lzo文件,然后上传到HDFS。因为kevinweil-hadoop-lzo提供了LzopOutpuStream,所以这一切很容易实现,需要注意的是该jar包依赖native code和lzo本身。本人试过Karmasphere的lzo-java项目,该项目用java重写了lzo压缩算法,但是该项目好久不更新,提问没人回,性能也只有原生lzo压缩性能一半,最难以忍受的是会有压缩后文件损坏的情况,果断放弃。

 

工具:lzo

地址:http://www.oberhumer.com/opensource/lzo/download/lzo-2.06.tar.gz 

         或者到 http://rpmfind.net 查找rpm包,下面两个是32bit redhat4和5的rpm包

          ftp://rpmfind.net/linux/dag/redhat/el4/en/i386/dag/RPMS/lzo-2.06-1.el4.rf.i386.rpm

  ftp://rpmfind.net/linux/dag/redhat/el5/en/i386/dag/RPMS/lzo-2.06-1.el5.rf.i386.rpm

 

工具:kevinweil-hadoop-lzo

地址:https://github.com/kevinweil/hadoop-lzo

 

 

-----------------------

 

1. 安装lzo和kevinweil-hadoop-lzo

   详见:http://heipark.iteye.com/blog/1172759

 

2. 设置环境变量

 

 

export JAVA_HOME=/usr/java/latest/
export LD_LIBRARY_PATH=/xxx/build/native/Linux-i386-32/lib

3. 写java代码

 

 

public class TestLzo {

	@Test
	public void test(String input, String output) throws IOException {

		BufferedReader textBr = new BufferedReader(new InputStreamReader(new FileInputStream(input)));
		int lzoBufferSize = 256 * 1024;
		LzoCompressor.CompressionStrategy strategy = LzoCompressor.CompressionStrategy.LZO1X_1;
		LzoCompressor lzoCompressor = new LzoCompressor(strategy, lzoBufferSize);

		LzopOutputStream lzopOut = new LzopOutputStream(new FileOutputStream(output), lzoCompressor, lzoBufferSize, strategy);
		String textLine;
		while ((textLine = textBr.readLine()) != null) {
			textLine += "\n";
			byte[] bytes = textLine.getBytes();
			lzopOut.write(bytes, 0, bytes.length);
		}
		textBr.close();
		lzopOut.close();
	}

	public static void main(String[] args) throws IOException {
		TestLzo lzo = new TestLzo();
		lzo.test(args[0], args[1]);
	}
}
 

4. 运行

 

/usr/java/latest/bin/java -cp .:hadoop-lzo-0.4.15.jar:hadoop-core-0.20.2-cdh3u4.jar:commons-logging-1.1.1.jar com/hadoop/compression/lzo/TestLzo abc.txt abc.txt.lzo  

 

--heipark

 

 

你可能感兴趣的:(hadoop lzo)