Search in sources :

Example 1 with BitmapInvertedIndexReader

use of com.linkedin.pinot.core.segment.index.readers.BitmapInvertedIndexReader in project pinot by linkedin.

the class ColumnIndexContainer method loadUnsorted.

private static ColumnIndexContainer loadUnsorted(String column, SegmentDirectory.Reader segmentReader, ColumnMetadata metadata, ImmutableDictionaryReader dictionary, boolean loadInverted) throws IOException {
    PinotDataBuffer fwdIndexBuffer = segmentReader.getIndexFor(column, ColumnIndexType.FORWARD_INDEX);
    SingleColumnSingleValueReader fwdIndexReader;
    if (dictionary != null) {
        fwdIndexReader = new FixedBitSingleValueReader(fwdIndexBuffer, metadata.getTotalDocs(), metadata.getBitsPerElement(), metadata.hasNulls());
    } else {
        // TODO: Replace hard-coded compressor with getting information from meta-data.
        fwdIndexReader = getRawIndexReader(fwdIndexBuffer, metadata.getDataType());
    }
    BitmapInvertedIndexReader invertedIndex = null;
    if (loadInverted) {
        PinotDataBuffer invertedIndexBuffer = segmentReader.getIndexFor(column, ColumnIndexType.INVERTED_INDEX);
        invertedIndex = new BitmapInvertedIndexReader(invertedIndexBuffer, metadata.getCardinality());
    }
    return new UnsortedSVColumnIndexContainer(column, metadata, fwdIndexReader, dictionary, invertedIndex);
}
Also used : SingleColumnSingleValueReader(com.linkedin.pinot.core.io.reader.SingleColumnSingleValueReader) PinotDataBuffer(com.linkedin.pinot.core.segment.memory.PinotDataBuffer) FixedBitSingleValueReader(com.linkedin.pinot.core.io.reader.impl.v1.FixedBitSingleValueReader) BitmapInvertedIndexReader(com.linkedin.pinot.core.segment.index.readers.BitmapInvertedIndexReader)

Example 2 with BitmapInvertedIndexReader

use of com.linkedin.pinot.core.segment.index.readers.BitmapInvertedIndexReader in project pinot by linkedin.

the class ColumnIndexContainer method loadMultiValue.

private static ColumnIndexContainer loadMultiValue(String column, SegmentDirectory.Reader segmentReader, ColumnMetadata metadata, ImmutableDictionaryReader dictionary, boolean loadInverted) throws IOException {
    PinotDataBuffer fwdIndexBuffer = segmentReader.getIndexFor(column, ColumnIndexType.FORWARD_INDEX);
    SingleColumnMultiValueReader<? extends ReaderContext> fwdIndexReader = new FixedBitMultiValueReader(fwdIndexBuffer, metadata.getTotalDocs(), metadata.getTotalNumberOfEntries(), metadata.getBitsPerElement(), false);
    BitmapInvertedIndexReader invertedIndex = null;
    if (loadInverted) {
        PinotDataBuffer invertedIndexBuffer = segmentReader.getIndexFor(column, ColumnIndexType.INVERTED_INDEX);
        invertedIndex = new BitmapInvertedIndexReader(invertedIndexBuffer, metadata.getCardinality());
    }
    return new UnSortedMVColumnIndexContainer(column, metadata, fwdIndexReader, dictionary, invertedIndex);
}
Also used : FixedBitMultiValueReader(com.linkedin.pinot.core.io.reader.impl.v1.FixedBitMultiValueReader) PinotDataBuffer(com.linkedin.pinot.core.segment.memory.PinotDataBuffer) BitmapInvertedIndexReader(com.linkedin.pinot.core.segment.index.readers.BitmapInvertedIndexReader)

Example 3 with BitmapInvertedIndexReader

use of com.linkedin.pinot.core.segment.index.readers.BitmapInvertedIndexReader in project pinot by linkedin.

the class BitmapInvertedIndexCreatorTest method validate.

private void validate(String colName, File bitmapIndexFile, int cardinality, Map<Integer, Set<Integer>> postingListMap) throws IOException {
    Assert.assertTrue(bitmapIndexFile.exists());
    PinotDataBuffer dataBuffer = PinotDataBuffer.fromFile(bitmapIndexFile, ReadMode.mmap, FileChannel.MapMode.READ_ONLY, "testing");
    BitmapInvertedIndexReader reader = new BitmapInvertedIndexReader(dataBuffer, cardinality);
    for (int i = 0; i < cardinality; i++) {
        ImmutableRoaringBitmap bitmap = reader.getImmutable(i);
        Set<Integer> expected = postingListMap.get(i);
        Assert.assertEquals(bitmap.getCardinality(), expected.size());
        int[] actual = bitmap.toArray();
        List<Integer> actualList = Ints.asList(actual);
        Assert.assertEquals(actualList, expected);
    }
    dataBuffer.close();
}
Also used : PinotDataBuffer(com.linkedin.pinot.core.segment.memory.PinotDataBuffer) ImmutableRoaringBitmap(org.roaringbitmap.buffer.ImmutableRoaringBitmap) BitmapInvertedIndexReader(com.linkedin.pinot.core.segment.index.readers.BitmapInvertedIndexReader)

Example 4 with BitmapInvertedIndexReader

use of com.linkedin.pinot.core.segment.index.readers.BitmapInvertedIndexReader in project pinot by linkedin.

the class BitmapPerformanceBenchmark method iterationSpeed.

public static void iterationSpeed(String indexSegmentDir, String column) throws Exception {
    File indexSegment = new File(indexSegmentDir);
    SegmentMetadataImpl segmentMetadata = new SegmentMetadataImpl(indexSegment);
    Map<String, BitmapInvertedIndexReader> bitMapIndexMap = new HashMap<String, BitmapInvertedIndexReader>();
    Map<String, Integer> cardinalityMap = new HashMap<String, Integer>();
    Map<String, ImmutableDictionaryReader> dictionaryMap = new HashMap<String, ImmutableDictionaryReader>();
    File bitMapIndexFile = new File(indexSegmentDir, column + ".bitmap.inv");
    ColumnMetadata columnMetadata = segmentMetadata.getColumnMetadataFor(column);
    int cardinality = columnMetadata.getCardinality();
    cardinalityMap.put(column, cardinality);
    PinotDataBuffer bitMapDataBuffer = PinotDataBuffer.fromFile(bitMapIndexFile, ReadMode.mmap, FileChannel.MapMode.READ_ONLY, "testing");
    BitmapInvertedIndexReader bitmapInvertedIndex = new BitmapInvertedIndexReader(bitMapDataBuffer, cardinality);
    File dictionaryFile = new File(indexSegmentDir + "/" + column + ".dict");
    SegmentDirectory segmentDirectory = SegmentDirectory.createFromLocalFS(indexSegment, segmentMetadata, ReadMode.mmap);
    SegmentDirectory.Reader segmentReader = segmentDirectory.createReader();
    ColumnIndexContainer container = ColumnIndexContainer.init(segmentReader, columnMetadata, null);
    ImmutableDictionaryReader dictionary = container.getDictionary();
    dictionaryMap.put(column, dictionary);
    // System.out.println(column + ":\t" + MemoryUtil.deepMemoryUsageOf(bitmapInvertedIndex));
    bitMapIndexMap.put(column, bitmapInvertedIndex);
    int dictId = dictionary.indexOf("na.us");
    ImmutableRoaringBitmap immutable = bitmapInvertedIndex.getImmutable(dictId);
    Iterator<Integer> iterator = immutable.iterator();
    int count = 0;
    long start = System.currentTimeMillis();
    while (iterator.hasNext()) {
        iterator.next();
        count = count + 1;
    }
    long end = System.currentTimeMillis();
    System.out.println(" matched: " + count + " Time to iterate:" + (end - start));
    bitMapDataBuffer.close();
}
Also used : ColumnMetadata(com.linkedin.pinot.core.segment.index.ColumnMetadata) HashMap(java.util.HashMap) ImmutableDictionaryReader(com.linkedin.pinot.core.segment.index.readers.ImmutableDictionaryReader) BitmapInvertedIndexReader(com.linkedin.pinot.core.segment.index.readers.BitmapInvertedIndexReader) SegmentDirectory(com.linkedin.pinot.core.segment.store.SegmentDirectory) ColumnIndexContainer(com.linkedin.pinot.core.segment.index.column.ColumnIndexContainer) PinotDataBuffer(com.linkedin.pinot.core.segment.memory.PinotDataBuffer) ImmutableRoaringBitmap(org.roaringbitmap.buffer.ImmutableRoaringBitmap) SegmentMetadataImpl(com.linkedin.pinot.core.segment.index.SegmentMetadataImpl) File(java.io.File)

Example 5 with BitmapInvertedIndexReader

use of com.linkedin.pinot.core.segment.index.readers.BitmapInvertedIndexReader in project pinot by linkedin.

the class BitmapPerformanceBenchmark method benchmarkIntersetionAndUnion.

public static void benchmarkIntersetionAndUnion(String indexSegmentDir) throws ConfigurationException, IOException, Exception {
    File[] listFiles = new File(indexSegmentDir).listFiles();
    File indexDir = new File(indexSegmentDir);
    SegmentMetadataImpl segmentMetadata = new SegmentMetadataImpl(indexDir);
    Map<String, BitmapInvertedIndexReader> bitMapIndexMap = new HashMap<String, BitmapInvertedIndexReader>();
    Map<String, Integer> cardinalityMap = new HashMap<String, Integer>();
    Map<String, ImmutableDictionaryReader> dictionaryMap = new HashMap<String, ImmutableDictionaryReader>();
    for (File file : listFiles) {
        if (!file.getName().endsWith("bitmap.inv")) {
            continue;
        }
        String column = file.getName().replaceAll(".bitmap.inv", "");
        ColumnMetadata columnMetadata = segmentMetadata.getColumnMetadataFor(column);
        int cardinality = columnMetadata.getCardinality();
        cardinalityMap.put(column, cardinality);
        System.out.println(column + "\t\t\t" + cardinality + "  \t" + columnMetadata.getDataType());
        PinotDataBuffer bitmapDataBuffer = PinotDataBuffer.fromFile(file, ReadMode.mmap, FileChannel.MapMode.READ_ONLY, "testing");
        BitmapInvertedIndexReader bitmapInvertedIndex = new BitmapInvertedIndexReader(bitmapDataBuffer, cardinality);
        File dictionaryFile = new File(indexSegmentDir + "/" + column + ".dict");
        SegmentDirectory segmentDirectory = SegmentDirectory.createFromLocalFS(indexDir, segmentMetadata, ReadMode.mmap);
        SegmentDirectory.Reader segmentReader = segmentDirectory.createReader();
        ColumnIndexContainer container = ColumnIndexContainer.init(segmentReader, columnMetadata, null);
        ImmutableDictionaryReader dictionary = container.getDictionary();
        if (columnMetadata.getDataType() == DataType.INT) {
            System.out.println("BitmapPerformanceBenchmark.main()");
            assert dictionary instanceof IntDictionary;
        }
        dictionaryMap.put(column, dictionary);
        // System.out.println(column + ":\t" + MemoryUtil.deepMemoryUsageOf(bitmapInvertedIndex));
        bitMapIndexMap.put(column, bitmapInvertedIndex);
        bitmapDataBuffer.close();
    }
    List<String> dimensionNamesList = segmentMetadata.getSchema().getDimensionNames();
    Collections.shuffle(dimensionNamesList);
    int NUM_TEST = 100;
    final int MAX_DIMENSIONS_PER_DIMENSION = 1;
    int MAX_DIMENSIONS_IN_WHERE_CLAUSE = 3;
    Random random = new Random();
    for (int numDimensions = 1; numDimensions <= MAX_DIMENSIONS_IN_WHERE_CLAUSE; numDimensions++) {
        for (int numValuesPerDimension = 1; numValuesPerDimension <= MAX_DIMENSIONS_PER_DIMENSION; numValuesPerDimension++) {
            int runCount = 0;
            while (runCount < NUM_TEST) {
                Collections.shuffle(dimensionNamesList);
                List<ImmutableRoaringBitmap> bitMaps = new ArrayList<ImmutableRoaringBitmap>();
                List<String> columnNameValuePairs = new ArrayList<String>();
                for (int i = 0; i < numDimensions; i++) {
                    String columnName = dimensionNamesList.get(i);
                    InvertedIndexReader bitmapInvertedIndex = bitMapIndexMap.get(columnName);
                    for (int j = 0; j < numValuesPerDimension; j++) {
                        int dictId = random.nextInt(cardinalityMap.get(columnName));
                        String dictValue = dictionaryMap.get(columnName).getStringValue(dictId);
                        columnNameValuePairs.add(columnName + ":" + dictValue);
                        ImmutableRoaringBitmap immutable = bitmapInvertedIndex.getImmutable(dictId);
                        bitMaps.add(immutable);
                    }
                }
                System.out.println("START**********************************");
                int[] cardinality = new int[bitMaps.size()];
                int[] sizes = new int[bitMaps.size()];
                for (int i = 0; i < bitMaps.size(); i++) {
                    ImmutableRoaringBitmap immutableRoaringBitmap = bitMaps.get(i);
                    cardinality[i] = immutableRoaringBitmap.getCardinality();
                    sizes[i] = immutableRoaringBitmap.getSizeInBytes();
                }
                System.out.println("\t#bitmaps:" + bitMaps.size());
                System.out.println("\tinput values:" + columnNameValuePairs);
                System.out.println("\tinput cardinality:" + Arrays.toString(cardinality));
                System.out.println("\tinput sizes:" + Arrays.toString(sizes));
                and(bitMaps);
                or(bitMaps);
                System.out.println("END**********************************");
                runCount = runCount + 1;
            }
        }
    }
}
Also used : ColumnMetadata(com.linkedin.pinot.core.segment.index.ColumnMetadata) HashMap(java.util.HashMap) ImmutableDictionaryReader(com.linkedin.pinot.core.segment.index.readers.ImmutableDictionaryReader) ArrayList(java.util.ArrayList) SegmentDirectory(com.linkedin.pinot.core.segment.store.SegmentDirectory) IntDictionary(com.linkedin.pinot.core.segment.index.readers.IntDictionary) Random(java.util.Random) ImmutableRoaringBitmap(org.roaringbitmap.buffer.ImmutableRoaringBitmap) SegmentMetadataImpl(com.linkedin.pinot.core.segment.index.SegmentMetadataImpl) BitmapInvertedIndexReader(com.linkedin.pinot.core.segment.index.readers.BitmapInvertedIndexReader) ColumnIndexContainer(com.linkedin.pinot.core.segment.index.column.ColumnIndexContainer) PinotDataBuffer(com.linkedin.pinot.core.segment.memory.PinotDataBuffer) BitmapInvertedIndexReader(com.linkedin.pinot.core.segment.index.readers.BitmapInvertedIndexReader) InvertedIndexReader(com.linkedin.pinot.core.segment.index.readers.InvertedIndexReader) File(java.io.File)

Aggregations

BitmapInvertedIndexReader (com.linkedin.pinot.core.segment.index.readers.BitmapInvertedIndexReader)5 PinotDataBuffer (com.linkedin.pinot.core.segment.memory.PinotDataBuffer)5 ImmutableRoaringBitmap (org.roaringbitmap.buffer.ImmutableRoaringBitmap)3 ColumnMetadata (com.linkedin.pinot.core.segment.index.ColumnMetadata)2 SegmentMetadataImpl (com.linkedin.pinot.core.segment.index.SegmentMetadataImpl)2 ColumnIndexContainer (com.linkedin.pinot.core.segment.index.column.ColumnIndexContainer)2 ImmutableDictionaryReader (com.linkedin.pinot.core.segment.index.readers.ImmutableDictionaryReader)2 SegmentDirectory (com.linkedin.pinot.core.segment.store.SegmentDirectory)2 File (java.io.File)2 HashMap (java.util.HashMap)2 SingleColumnSingleValueReader (com.linkedin.pinot.core.io.reader.SingleColumnSingleValueReader)1 FixedBitMultiValueReader (com.linkedin.pinot.core.io.reader.impl.v1.FixedBitMultiValueReader)1 FixedBitSingleValueReader (com.linkedin.pinot.core.io.reader.impl.v1.FixedBitSingleValueReader)1 IntDictionary (com.linkedin.pinot.core.segment.index.readers.IntDictionary)1 InvertedIndexReader (com.linkedin.pinot.core.segment.index.readers.InvertedIndexReader)1 ArrayList (java.util.ArrayList)1 Random (java.util.Random)1