org.apache.hadoop.hbase.ByteBufferExtendedCell Maven / Gradle / Ivy
/*
* 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
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* 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.hadoop.hbase;
import java.nio.ByteBuffer;
import org.apache.yetus.audience.InterfaceAudience;
/**
* This class is a server side extension to the {@link Cell} interface. It is used when the Cell is
* backed by a {@link ByteBuffer}: i.e. cell instanceof ByteBufferedCell.
*
* This class has getters for the row, column family, column qualifier, value and tags hosting
* ByteBuffers. It also has getters of the *position* within a ByteBuffer where these field bytes
* begin. These are needed because a single ByteBuffer may back one or many Cell instances -- it
* depends on the implementation -- so the ByteBuffer position as returned by
* {@link ByteBuffer#arrayOffset()} cannot be relied upon. Also, do not confuse these position
* methods with the getXXXOffset methods from the super Interface, {@link Cell}; dependent up on
* implementation, the Cell getXXXOffset methods can return the same value as a call to its
* equivalent position method from below BUT they can also stray; if a ByteBufferedCell, use the
* below position methods to find where a field begins.
*
* Use the getXXXLength methods from Cell to find a fields length.
*
* A Cell object can be of this type only on the server side.
*
* WARNING: If a Cell is backed by an offheap ByteBuffer, any call to getXXXArray() will result in a
* temporary byte array creation and a bytes copy. Avoid these allocations by using the appropriate
* Cell access server-side: i.e. ByteBufferedCell when backed by a ByteBuffer and Cell when it is
* not.
*/
/*
* Even though all the methods are abstract, ByteBufferExtendedCell is not made to be an interface
* with intent. In CellComparator compare method, we have instance of check to decide whether to use
* getXXXArray() or getXXXByteBuffer(). This is a very hot method in read and write paths. if (left
* instanceof ByteBufferExtendedCell && right instanceof ByteBufferExtendedCell) { .... } if (left
* instanceof ByteBufferExtendedCell) { .... } if (right instanceof ByteBufferExtendedCell) { .... }
* return Bytes.compareTo(left.getRowArray(), left.getRowOffset(), left.getRowLength(),
* right.getRowArray(), right.getRowOffset(), right.getRowLength()); We did JMH micro benchmark
* tests with both left and right cells as ByteBufferExtendedCell, one only ByteBufferExtendedCell
* and both as Cells. This is compared against JMH results on compare logic with out any instance of
* checks. We noticed that if ByteBufferExtendedCell is an interface, the benchmark result seems to
* be very bad for case of both right and left are Cell only (Not ByteBufferExtendedCell). When
* ByteBufferExtendedCell is an abstract class all 4 possible cases giving almost similar
* performance number compared with compare logic with no instance of checks.
*/
@InterfaceAudience.Private
public abstract class ByteBufferExtendedCell implements ExtendedCell {
/**
* @return The {@link ByteBuffer} containing the row bytes.
*/
public abstract ByteBuffer getRowByteBuffer();
/**
* @return Position in the {@link ByteBuffer} where row bytes start
*/
public abstract int getRowPosition();
/**
* @return The {@link ByteBuffer} containing the column family bytes.
*/
public abstract ByteBuffer getFamilyByteBuffer();
/**
* @return Position in the {@link ByteBuffer} where column family bytes start
*/
public abstract int getFamilyPosition();
/**
* @return The {@link ByteBuffer} containing the column qualifier bytes.
*/
public abstract ByteBuffer getQualifierByteBuffer();
/**
* @return Position in the {@link ByteBuffer} where column qualifier bytes start
*/
public abstract int getQualifierPosition();
/**
* @return The {@link ByteBuffer} containing the value bytes.
*/
public abstract ByteBuffer getValueByteBuffer();
/**
* @return Position in the {@link ByteBuffer} where value bytes start
*/
public abstract int getValuePosition();
/**
* @return The {@link ByteBuffer} containing the tag bytes.
*/
public abstract ByteBuffer getTagsByteBuffer();
/**
* @return Position in the {@link ByteBuffer} where tag bytes start
*/
public abstract int getTagsPosition();
}