ZOOKEEPER-2870: Improve the efficiency of AtomicFileOutputStream

Author: Fangmin Lyu <allenlyu@fb.com>

Reviewers: Michael Han <hanm@apache.org>

Closes #331 from lvfangmin/ZOOKEEPER-2870

src/java/main/org/apache/zookeeper/common/AtomicFileOutputStream.java

@@ -35,10 +35,10 @@ import org.slf4j.LoggerFactory;
  * A FileOutputStream that has the property that it will only show up at its
  * destination once it has been entirely written and flushed to disk. While
  * being written, it will use a .tmp suffix.
- * 
+ *
  * When the output stream is closed, it is flushed, fsynced, and will be moved
  * into place, overwriting any file that already exists at that location.
- * 
+ *
  * <b>NOTE</b>: on Windows platforms, it will not atomically replace the target
  * file - instead the target file is deleted before this one is moved into
  * place.
@@ -63,6 +63,17 @@ public class AtomicFileOutputStream extends FilterOutputStream {
                 .getAbsoluteFile();
     }
 
+    /**
+     * The default write method in FilterOutputStream does not call the write
+     * method of its underlying input stream with the same arguments. Instead
+     * it writes the data byte by byte, override it here to make it more
+     * efficient.
+     */
+    @Override
+    public void write(byte b[], int off, int len) throws IOException {
+        out.write(b, off, len);
+    }
+
     @Override
     public void close() throws IOException {
         boolean triedToClose = false, success = false;