use of com.linkedin.pinot.core.segment.memory.PinotDataBuffer in project pinot by linkedin.
the class FixedByteWidthRowColDataFileWriterTest method testSingleColInt.
@Test
public void testSingleColInt() throws Exception {
File file = new File("test_single_col_writer.dat");
file.delete();
int rows = 100;
int cols = 1;
int[] columnSizes = new int[] { 4 };
FixedByteSingleValueMultiColWriter writer = new FixedByteSingleValueMultiColWriter(file, rows, cols, columnSizes);
int[] data = new int[rows];
Random r = new Random();
for (int i = 0; i < rows; i++) {
data[i] = r.nextInt();
writer.setInt(i, 0, data[i]);
}
writer.close();
File rfile = new File("test_single_col_writer.dat");
PinotDataBuffer buffer = PinotDataBuffer.fromFile(rfile, ReadMode.mmap, FileChannel.MapMode.READ_WRITE, "testing");
FixedByteSingleValueMultiColReader reader = new FixedByteSingleValueMultiColReader(buffer, rows, cols, columnSizes);
for (int i = 0; i < rows; i++) {
Assert.assertEquals(reader.getInt(i, 0), data[i]);
}
reader.close();
rfile.delete();
}
use of com.linkedin.pinot.core.segment.memory.PinotDataBuffer 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();
}
use of com.linkedin.pinot.core.segment.memory.PinotDataBuffer in project pinot by linkedin.
the class BenchmarkFileRead method readSVs.
/*@Benchmark
@OutputTimeUnit(TimeUnit.MILLISECONDS)
public void test() {
byteBuffer.rewind();
byte[] rawData = new byte[length];
byteBuffer.rewind();
for (int i = 0; i < length; i++) {
rawData[i] = byteBuffer.get();
}
}*/
@Benchmark
@BenchmarkMode({ Mode.SampleTime })
@OutputTimeUnit(TimeUnit.MILLISECONDS)
public void readSVs() throws IOException {
int rows = 25000000;
int columnSizeInBits = 3;
boolean isMMap = true;
boolean hasNulls = false;
PinotDataBuffer dataBuffer = PinotDataBuffer.fromFile(file, ReadMode.mmap, FileChannel.MapMode.READ_ONLY, "benchmark");
FixedBitSingleValueReader reader = new FixedBitSingleValueReader(dataBuffer, rows, columnSizeInBits, hasNulls);
int[] result2 = new int[rows];
for (int i = 0; i < rows; i++) {
result2[i] = reader.getInt(i);
}
}
use of com.linkedin.pinot.core.segment.memory.PinotDataBuffer 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();
}
use of com.linkedin.pinot.core.segment.memory.PinotDataBuffer 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;
}
}
}
}
Aggregations