001/*
002 * Licensed to the Apache Software Foundation (ASF) under one
003 * or more contributor license agreements.  See the NOTICE file
004 * distributed with this work for additional information
005 * regarding copyright ownership.  The ASF licenses this file
006 * to you under the Apache License, Version 2.0 (the
007 * "License"); you may not use this file except in compliance
008 * with the License.  You may obtain a copy of the License at
009 *
010 *     http://www.apache.org/licenses/LICENSE-2.0
011 *
012 * Unless required by applicable law or agreed to in writing, software
013 * distributed under the License is distributed on an "AS IS" BASIS,
014 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
015 * See the License for the specific language governing permissions and
016 * limitations under the License.
017 */
018package org.apache.hadoop.hbase.client;
019
020import java.io.Closeable;
021import java.io.IOException;
022import java.util.Collections;
023import java.util.List;
024import java.util.Map;
025import org.apache.hadoop.conf.Configuration;
026import org.apache.hadoop.hbase.TableName;
027import org.apache.yetus.audience.InterfaceAudience;
028
029/**
030 * <p>
031 * Used to communicate with a single HBase table similar to {@link Table} but meant for batched,
032 * asynchronous puts. Obtain an instance from a {@link Connection} and call {@link #close()}
033 * afterwards. Customizations can be applied to the {@code BufferedMutator} via the
034 * {@link BufferedMutatorParams}.
035 * </p>
036 * <p>
037 * Exception handling with asynchronously via the {@link BufferedMutator.ExceptionListener}. The
038 * default implementation is to throw the exception upon receipt. This behavior can be overridden
039 * with a custom implementation, provided as a parameter with
040 * {@link BufferedMutatorParams#listener(BufferedMutator.ExceptionListener)}.
041 * </p>
042 * <p>
043 * Map/Reduce jobs are good use cases for using {@code BufferedMutator}. Map/reduce jobs benefit
044 * from batching, but have no natural flush point. {@code BufferedMutator} receives the puts from
045 * the M/R job and will batch puts based on some heuristic, such as the accumulated size of the
046 * puts, and submit batches of puts asynchronously so that the M/R logic can continue without
047 * interruption.
048 * </p>
049 * <p>
050 * {@code BufferedMutator} can also be used on more exotic circumstances. Map/Reduce batch jobs will
051 * have a single {@code BufferedMutator} per thread. A single {@code BufferedMutator} can also be
052 * effectively used in high volume online systems to batch puts, with the caveat that extreme
053 * circumstances, such as JVM or machine failure, may cause some data loss.
054 * </p>
055 * <p>
056 * NOTE: This class replaces the functionality that used to be available via
057 * HTable#setAutoFlush(boolean) set to {@code false}.
058 * </p>
059 * <p>
060 * See also the {@code BufferedMutatorExample} in the hbase-examples module.
061 * </p>
062 * @see ConnectionFactory
063 * @see Connection
064 * @since 1.0.0
065 */
066@InterfaceAudience.Public
067public interface BufferedMutator extends Closeable {
068  /**
069   * Key to use setting non-default BufferedMutator implementation in Configuration.
070   */
071  String CLASSNAME_KEY = "hbase.client.bufferedmutator.classname";
072
073  /**
074   * Having the timer tick run more often that once every 100ms is needless and will probably cause
075   * too many timer events firing having a negative impact on performance.
076   */
077  long MIN_WRITE_BUFFER_PERIODIC_FLUSH_TIMERTICK_MS = 100;
078
079  /**
080   * Gets the fully qualified table name instance of the table that this BufferedMutator writes to.
081   */
082  TableName getName();
083
084  /**
085   * Returns the {@link org.apache.hadoop.conf.Configuration} object used by this instance.
086   * <p>
087   * The reference returned is not a copy, so any change made to it will affect this instance.
088   */
089  Configuration getConfiguration();
090
091  /**
092   * Sends a {@link Mutation} to the table. The mutations will be buffered and sent over the wire as
093   * part of a batch. Currently only supports {@link Put} and {@link Delete} mutations.
094   * @param mutation The data to send.
095   * @throws IOException if a remote or network exception occurs.
096   */
097  void mutate(Mutation mutation) throws IOException;
098
099  /**
100   * Send some {@link Mutation}s to the table. The mutations will be buffered and sent over the wire
101   * as part of a batch. There is no guarantee of sending entire content of {@code mutations} in a
102   * single batch; it will be broken up according to the write buffer capacity.
103   * @param mutations The data to send.
104   * @throws IOException if a remote or network exception occurs.
105   */
106  void mutate(List<? extends Mutation> mutations) throws IOException;
107
108  /**
109   * Performs a {@link #flush()} and releases any resources held.
110   * @throws IOException if a remote or network exception occurs.
111   */
112  @Override
113  void close() throws IOException;
114
115  /**
116   * Executes all the buffered, asynchronous {@link Mutation} operations and waits until they are
117   * done.
118   * @throws IOException if a remote or network exception occurs.
119   */
120  void flush() throws IOException;
121
122  /**
123   * Sets the maximum time before the buffer is automatically flushed checking once per second.
124   * @param timeoutMs The maximum number of milliseconds how long records may be buffered before
125   *                  they are flushed. Set to 0 to disable.
126   */
127  default void setWriteBufferPeriodicFlush(long timeoutMs) {
128    setWriteBufferPeriodicFlush(timeoutMs, 1000L);
129  }
130
131  /**
132   * Sets the maximum time before the buffer is automatically flushed.
133   * @param timeoutMs   The maximum number of milliseconds how long records may be buffered before
134   *                    they are flushed. Set to 0 to disable.
135   * @param timerTickMs The number of milliseconds between each check if the timeout has been
136   *                    exceeded. Must be 100ms (as defined in
137   *                    {@link #MIN_WRITE_BUFFER_PERIODIC_FLUSH_TIMERTICK_MS}) or larger to avoid
138   *                    performance problems.
139   */
140  default void setWriteBufferPeriodicFlush(long timeoutMs, long timerTickMs) {
141    throw new UnsupportedOperationException(
142      "The BufferedMutator::setWriteBufferPeriodicFlush has not been implemented");
143  }
144
145  /**
146   * Disable periodic flushing of the write buffer.
147   */
148  default void disableWriteBufferPeriodicFlush() {
149    setWriteBufferPeriodicFlush(0, MIN_WRITE_BUFFER_PERIODIC_FLUSH_TIMERTICK_MS);
150  }
151
152  /**
153   * Returns the current periodic flush timeout value in milliseconds.
154   * @return The maximum number of milliseconds how long records may be buffered before they are
155   *         flushed. The value 0 means this is disabled.
156   */
157  default long getWriteBufferPeriodicFlushTimeoutMs() {
158    throw new UnsupportedOperationException(
159      "The BufferedMutator::getWriteBufferPeriodicFlushTimeoutMs has not been implemented");
160  }
161
162  /**
163   * Returns the current periodic flush timertick interval in milliseconds.
164   * @return The number of milliseconds between each check if the timeout has been exceeded. This
165   *         value only has a real meaning if the timeout has been set to > 0
166   */
167  default long getWriteBufferPeriodicFlushTimerTickMs() {
168    throw new UnsupportedOperationException(
169      "The BufferedMutator::getWriteBufferPeriodicFlushTimerTickMs has not been implemented");
170  }
171
172  /**
173   * Returns the maximum size in bytes of the write buffer for this HTable.
174   * <p>
175   * The default value comes from the configuration parameter {@code hbase.client.write.buffer}.
176   * @return The size of the write buffer in bytes.
177   */
178  default long getWriteBufferSize() {
179    throw new UnsupportedOperationException(
180      "The BufferedMutator::getWriteBufferSize has not been implemented");
181  }
182
183  /**
184   * Set rpc timeout for this mutator instance
185   */
186  default void setRpcTimeout(int timeout) {
187    throw new UnsupportedOperationException(
188      "The BufferedMutator::setRpcTimeout has not been implemented");
189  }
190
191  /**
192   * Set operation timeout for this mutator instance
193   */
194  default void setOperationTimeout(int timeout) {
195    throw new UnsupportedOperationException(
196      "The BufferedMutator::setOperationTimeout has not been implemented");
197  }
198
199  /**
200   * Returns the rpc request attributes.
201   */
202  default Map<String, byte[]> getRequestAttributes() {
203    return Collections.emptyMap();
204  }
205
206  /**
207   * The maximum number of mutations that this buffered mutator will buffer before flushing them
208   */
209  default int getMaxMutations() {
210    throw new UnsupportedOperationException(
211      "The BufferedMutator::getMaxMutations has not been implemented");
212  }
213
214  /**
215   * Listens for asynchronous exceptions on a {@link BufferedMutator}.
216   */
217  @InterfaceAudience.Public
218  interface ExceptionListener {
219    public void onException(RetriesExhaustedWithDetailsException exception, BufferedMutator mutator)
220      throws RetriesExhaustedWithDetailsException;
221  }
222}