org.apache.hadoop.mapreduce.checkpoint.CheckpointService 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.mapreduce.checkpoint;

import java.io.IOException;
import java.nio.channels.ReadableByteChannel;
import java.nio.channels.WritableByteChannel;

/**
 * The CheckpointService provides a simple API to store and retrieve the state of a task.
 *
 * Checkpoints are atomic, single-writer, write-once, multiple-readers,
 * ready-many type of objects. This is provided by releasing the CheckpointID
 * for a checkpoint only upon commit of the checkpoint, and by preventing a
 * checkpoint to be re-opened for writes.
 *
 * Non-functional properties such as durability, availability, compression,
 * garbage collection, quotas are left to the implementation.
 *
 * This API is envisioned as the basic building block for a checkpoint service,
 * on top of which richer interfaces can be layered (e.g., frameworks providing
 * object-serialization, checkpoint metadata and provenance, etc.)
 *
 */
public interface CheckpointService {

  public interface CheckpointWriteChannel extends WritableByteChannel { }
  public interface CheckpointReadChannel extends ReadableByteChannel { }

  /**
   * This method creates a checkpoint and provide a channel to write to it. The
   * name/location of the checkpoint are unknown to the user as of this time, in
   * fact, the CheckpointID is not released to the user until commit is called.
   * This makes enforcing atomicity of writes easy.
   * @return a channel that can be used to write to the checkpoint
   * @throws IOException
   * @throws InterruptedException
   */
  public CheckpointWriteChannel create()
    throws IOException, InterruptedException;

  /**
   * Used to finalize and existing checkpoint. It returns the CheckpointID that
   * can be later used to access (read-only) this checkpoint. This guarantees
   * atomicity of the checkpoint.
   * @param ch the CheckpointWriteChannel to commit
   * @return a CheckpointID
   * @throws IOException
   * @throws InterruptedException
   */
  public CheckpointID commit(CheckpointWriteChannel ch)
    throws IOException, InterruptedException;

  /**
   * Dual to commit, it aborts the current checkpoint. Garbage collection
   * choices are left to the implementation. The CheckpointID is not generated
   * nor released to the user so the checkpoint is not accessible.
   * @param ch the CheckpointWriteChannel to abort
   * @throws IOException
   * @throws InterruptedException
   */
  public void abort(CheckpointWriteChannel ch)
      throws IOException, InterruptedException;

  /**
   * Given a CheckpointID returns a reading channel.
   * @param id CheckpointID for the checkpoint to be opened
   * @return a CheckpointReadChannel
   * @throws IOException
   * @throws InterruptedException
   */
  public CheckpointReadChannel open(CheckpointID id)
    throws IOException, InterruptedException;

  /**
   * It discards an existing checkpoint identified by its CheckpointID.
   * @param  id CheckpointID for the checkpoint to be deleted
   * @return a boolean confirming success of the deletion
   * @throws IOException
   * @throws InterruptedException
   */
  public boolean delete(CheckpointID id)
    throws IOException, InterruptedException;

}