drill-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From parthchandra <...@git.apache.org>
Subject [GitHub] drill pull request #611: Drill-4800: Improve parquet reader performance
Date Fri, 04 Nov 2016 17:53:59 GMT
Github user parthchandra commented on a diff in the pull request:

    --- Diff: exec/java-exec/src/main/java/org/apache/drill/exec/util/filereader/BufferedDirectBufInputStream.java
    @@ -0,0 +1,460 @@
    + * Licensed to the Apache Software Foundation (ASF) under one
    + * or more contributor license agreements.  See the NOTICE file
    + * distributed with this work for additional information
    + * regarding copyright ownership.  The ASF licenses this file
    + * to you under the Apache License, Version 2.0 (the
    + * "License"); you may not use this file except in compliance
    + * with the License.  You may obtain a copy of the License at
    + * <p/>
    + * http://www.apache.org/licenses/LICENSE-2.0
    + * <p/>
    + * Unless required by applicable law or agreed to in writing, software
    + * distributed under the License is distributed on an "AS IS" BASIS,
    + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    + * See the License for the specific language governing permissions and
    + * limitations under the License.
    + */
    +package org.apache.drill.exec.util.filereader;
    +import com.google.common.base.Preconditions;
    +import io.netty.buffer.DrillBuf;
    +import org.apache.drill.common.config.DrillConfig;
    +import org.apache.drill.exec.memory.BufferAllocator;
    +import org.apache.drill.exec.memory.RootAllocatorFactory;
    +import org.apache.hadoop.conf.Configuration;
    +import org.apache.hadoop.fs.ByteBufferReadable;
    +import org.apache.hadoop.fs.FSDataInputStream;
    +import org.apache.hadoop.fs.FileSystem;
    +import org.apache.hadoop.fs.Path;
    +import org.apache.parquet.hadoop.Footer;
    +import org.apache.parquet.hadoop.ParquetFileReader;
    +import org.apache.parquet.hadoop.metadata.BlockMetaData;
    +import org.apache.parquet.hadoop.metadata.ColumnChunkMetaData;
    +import org.apache.parquet.hadoop.util.CompatibilityUtil;
    +import java.io.Closeable;
    +import java.io.IOException;
    +import java.io.InputStream;
    +import java.nio.ByteBuffer;
    +import java.util.List;
    + * <code>BufferedDirectBufInputStream</code>  reads from the
    + * underlying <code>InputStream</code> in blocks of data, into an
    + * internal buffer. The internal buffer is a direct memory backed
    + * buffer. The implementation is similar to the <code>BufferedInputStream</code>
    + * class except that the internal buffer is a Drillbuf and
    + * not a byte array. The mark and reset methods of the underlying
    + * <code>InputStream</code>are not supported.
    --- End diff --
    That's a question that's been asked of Drill before. The Parquet reader at one point did
read from the file system using byte arrays, but the overhead of copying from byte arrays
to direct memory was (it turned out) expensive. It also increased the heap usage. 
    Given that, we switched to reading directly into DirectByteBuffers which did improve performance
somewhat, and definitely reduced the heap memory requirements.
    I couldn't think of any other way to read from the FS into direct memory without writing
this class. Using BufferedStream would have taken us back to where we were before.

If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.

View raw message