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.regionserver;
019
020import java.io.IOException;
021import java.util.ArrayList;
022import java.util.List;
023import java.util.Random;
024
025import org.apache.hadoop.hbase.CompareOperator;
026import org.apache.hadoop.hbase.DoNotRetryIOException;
027import org.apache.hadoop.hbase.HBaseClassTestRule;
028import org.apache.hadoop.hbase.HBaseTestingUtility;
029import org.apache.hadoop.hbase.HColumnDescriptor;
030import org.apache.hadoop.hbase.HTableDescriptor;
031import org.apache.hadoop.hbase.TableName;
032import org.apache.hadoop.hbase.client.Admin;
033import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;
034import org.apache.hadoop.hbase.client.Connection;
035import org.apache.hadoop.hbase.client.Put;
036import org.apache.hadoop.hbase.client.Result;
037import org.apache.hadoop.hbase.client.ResultScanner;
038import org.apache.hadoop.hbase.client.Scan;
039import org.apache.hadoop.hbase.client.Table;
040import org.apache.hadoop.hbase.client.TableDescriptor;
041import org.apache.hadoop.hbase.client.TableDescriptorBuilder;
042import org.apache.hadoop.hbase.filter.CompareFilter;
043import org.apache.hadoop.hbase.filter.SingleColumnValueFilter;
044import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;
045import org.apache.hadoop.hbase.testclassification.LargeTests;
046import org.apache.hadoop.hbase.testclassification.RegionServerTests;
047import org.apache.hadoop.hbase.util.Bytes;
048import org.junit.AfterClass;
049import org.junit.BeforeClass;
050import org.junit.ClassRule;
051import org.junit.Rule;
052import org.junit.Test;
053import org.junit.experimental.categories.Category;
054import org.junit.rules.TestName;
055import org.slf4j.Logger;
056import org.slf4j.LoggerFactory;
057import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;
058import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLineParser;
059import org.apache.hbase.thirdparty.org.apache.commons.cli.GnuParser;
060import org.apache.hbase.thirdparty.org.apache.commons.cli.HelpFormatter;
061import org.apache.hbase.thirdparty.org.apache.commons.cli.Option;
062import org.apache.hbase.thirdparty.org.apache.commons.cli.Options;
063
064/**
065 * Test performance improvement of joined scanners optimization:
066 * https://issues.apache.org/jira/browse/HBASE-5416
067 */
068@Category({RegionServerTests.class, LargeTests.class})
069public class TestJoinedScanners {
070
071  @ClassRule
072  public static final HBaseClassTestRule CLASS_RULE =
073      HBaseClassTestRule.forClass(TestJoinedScanners.class);
074
075  private static final Logger LOG = LoggerFactory.getLogger(TestJoinedScanners.class);
076
077  private static final HBaseTestingUtility TEST_UTIL = new HBaseTestingUtility();
078
079  private static final byte[] cf_essential = Bytes.toBytes("essential");
080  private static final byte[] cf_joined = Bytes.toBytes("joined");
081  private static final byte[] col_name = Bytes.toBytes("a");
082  private static final byte[] flag_yes = Bytes.toBytes("Y");
083  private static final byte[] flag_no = Bytes.toBytes("N");
084
085  private static DataBlockEncoding blockEncoding = DataBlockEncoding.FAST_DIFF;
086  private static int selectionRatio = 30;
087  private static int valueWidth = 128 * 1024;
088
089  @Rule
090  public TestName name = new TestName();
091
092  @BeforeClass
093  public static void setUpBeforeClass() throws Exception {
094    final int DEFAULT_BLOCK_SIZE = 1024 * 1024;
095    TEST_UTIL.getConfiguration().setLong("dfs.blocksize", DEFAULT_BLOCK_SIZE);
096    TEST_UTIL.getConfiguration().setInt("dfs.replication", 1);
097    TEST_UTIL.getConfiguration().setLong("hbase.hregion.max.filesize", 322122547200L);
098
099    String[] dataNodeHosts = new String[] {"host1", "host2", "host3"};
100    int regionServersCount = 3;
101    TEST_UTIL.startMiniCluster(1, regionServersCount, dataNodeHosts);
102  }
103
104  @AfterClass
105  public static void tearDownAfterClass() throws Exception {
106    TEST_UTIL.shutdownMiniCluster();
107  }
108
109  @Test
110  public void testJoinedScanners() throws Exception {
111    byte[][] families = {cf_essential, cf_joined};
112
113    final TableName tableName = TableName.valueOf(name.getMethodName());
114    HTableDescriptor desc = new HTableDescriptor(tableName);
115    for (byte[] family : families) {
116      HColumnDescriptor hcd = new HColumnDescriptor(family);
117      hcd.setDataBlockEncoding(blockEncoding);
118      desc.addFamily(hcd);
119    }
120    TEST_UTIL.getAdmin().createTable(desc);
121    Table ht = TEST_UTIL.getConnection().getTable(tableName);
122
123    long rows_to_insert = 1000;
124    int insert_batch = 20;
125    long time = System.nanoTime();
126    Random rand = new Random(time);
127
128    LOG.info("Make " + Long.toString(rows_to_insert) + " rows, total size = " + Float
129      .toString(rows_to_insert * valueWidth / 1024 / 1024) + " MB");
130
131    byte[] val_large = new byte[valueWidth];
132
133    List<Put> puts = new ArrayList<>();
134
135    for (long i = 0; i < rows_to_insert; i++) {
136      Put put = new Put(Bytes.toBytes(Long.toString(i)));
137      if (rand.nextInt(100) <= selectionRatio) {
138        put.addColumn(cf_essential, col_name, flag_yes);
139      } else {
140        put.addColumn(cf_essential, col_name, flag_no);
141      }
142      put.addColumn(cf_joined, col_name, val_large);
143      puts.add(put);
144      if (puts.size() >= insert_batch) {
145        ht.put(puts);
146        puts.clear();
147      }
148    }
149    if (!puts.isEmpty()) {
150      ht.put(puts);
151      puts.clear();
152    }
153
154    LOG.info("Data generated in "
155      + Double.toString((System.nanoTime() - time) / 1000000000.0) + " seconds");
156
157    boolean slow = true;
158    for (int i = 0; i < 10; ++i) {
159      runScanner(ht, slow);
160      slow = !slow;
161    }
162
163    ht.close();
164  }
165
166  private void runScanner(Table table, boolean slow) throws Exception {
167    long time = System.nanoTime();
168    Scan scan = new Scan();
169    scan.addColumn(cf_essential, col_name);
170    scan.addColumn(cf_joined, col_name);
171
172    SingleColumnValueFilter filter = new SingleColumnValueFilter(
173        cf_essential, col_name, CompareFilter.CompareOp.EQUAL, flag_yes);
174    filter.setFilterIfMissing(true);
175    scan.setFilter(filter);
176    scan.setLoadColumnFamiliesOnDemand(!slow);
177
178    ResultScanner result_scanner = table.getScanner(scan);
179    Result res;
180    long rows_count = 0;
181    while ((res = result_scanner.next()) != null) {
182      rows_count++;
183    }
184
185    double timeSec = (System.nanoTime() - time) / 1000000000.0;
186    result_scanner.close();
187    LOG.info((slow ? "Slow" : "Joined") + " scanner finished in " + Double.toString(timeSec)
188      + " seconds, got " + Long.toString(rows_count/2) + " rows");
189  }
190
191  private static Options options = new Options();
192
193  /**
194   * Command line interface:
195   * @param args
196   * @throws IOException if there is a bug while reading from disk
197   */
198  public static void main(final String[] args) throws Exception {
199    Option encodingOption = new Option("e", "blockEncoding", true,
200      "Data block encoding; Default: FAST_DIFF");
201    encodingOption.setRequired(false);
202    options.addOption(encodingOption);
203
204    Option ratioOption = new Option("r", "selectionRatio", true,
205      "Ratio of selected rows using essential column family");
206    ratioOption.setRequired(false);
207    options.addOption(ratioOption);
208
209    Option widthOption = new Option("w", "valueWidth", true,
210      "Width of value for non-essential column family");
211    widthOption.setRequired(false);
212    options.addOption(widthOption);
213
214    CommandLineParser parser = new GnuParser();
215    CommandLine cmd = parser.parse(options, args);
216    if (args.length < 1) {
217      HelpFormatter formatter = new HelpFormatter();
218      formatter.printHelp("TestJoinedScanners", options, true);
219    }
220
221    if (cmd.hasOption("e")) {
222      blockEncoding = DataBlockEncoding.valueOf(cmd.getOptionValue("e"));
223    }
224    if (cmd.hasOption("r")) {
225      selectionRatio = Integer.parseInt(cmd.getOptionValue("r"));
226    }
227    if (cmd.hasOption("w")) {
228      valueWidth = Integer.parseInt(cmd.getOptionValue("w"));
229    }
230    // run the test
231    TestJoinedScanners test = new TestJoinedScanners();
232    test.testJoinedScanners();
233  }
234
235  @Test(expected = DoNotRetryIOException.class)
236  public void testWithReverseScan() throws Exception {
237    try (Connection con = TEST_UTIL.getConnection(); Admin admin = con.getAdmin()) {
238      TableName tableName = TableName.valueOf(name.getMethodName());
239
240      TableDescriptor tableDescriptor = TableDescriptorBuilder.newBuilder(tableName)
241          .setColumnFamily(ColumnFamilyDescriptorBuilder.of("cf1"))
242          .setColumnFamily(ColumnFamilyDescriptorBuilder.of("cf2"))
243          .build();
244      admin.createTable(tableDescriptor);
245
246      try (Table table = con.getTable(tableName)) {
247        SingleColumnValueFilter filter = new SingleColumnValueFilter(Bytes.toBytes("cf1"),
248          Bytes.toBytes("col"), CompareOperator.EQUAL, Bytes.toBytes("val"));
249        filter.setFilterIfMissing(true);
250
251        // Reverse scan with loading CFs on demand
252        Scan scan = new Scan();
253        scan.setFilter(filter);
254        scan.setReversed(true);
255        scan.setLoadColumnFamiliesOnDemand(true);
256
257        try (ResultScanner scanner = table.getScanner(scan)) {
258          // DoNotRetryIOException should occur
259          scanner.next();
260        }
261      }
262    }
263  }
264}