add compressed variable-size ints column type

This commit is contained in:
Xavier Léauté 2015-04-13 21:57:19 -07:00
parent ce928d9636
commit d20128b89b
3 changed files with 837 additions and 0 deletions

View File

@ -0,0 +1,71 @@
/*
* Licensed to Metamarkets Group Inc. (Metamarkets) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. Metamarkets 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 io.druid.segment.data;
import com.google.common.collect.Ordering;
import java.nio.ByteBuffer;
import java.nio.ByteOrder;
public class CompressedByteBufferObjectStrategy extends FixedSizeCompressedObjectStrategy<ByteBuffer>
{
public static final Ordering<Comparable> ORDERING = Ordering.natural().nullsFirst();
public static CompressedByteBufferObjectStrategy getBufferForOrder(final ByteOrder order, final CompressionStrategy compression, final int sizePer)
{
return new CompressedByteBufferObjectStrategy(order, compression, sizePer);
}
public CompressedByteBufferObjectStrategy(
ByteOrder order,
CompressionStrategy compression,
final int sizePer
)
{
super(
order, new BufferConverter<ByteBuffer>()
{
@Override
public ByteBuffer convert(ByteBuffer buf)
{
return buf;
}
@Override
public int compare(ByteBuffer lhs, ByteBuffer rhs)
{
return ORDERING.compare(lhs, rhs);
}
@Override
public int sizeOf(int count)
{
return count; // 1 byte per element
}
@Override
public ByteBuffer combine(ByteBuffer into, ByteBuffer from)
{
return into.put(from);
}
}, compression, sizePer
);
}
}

View File

@ -0,0 +1,395 @@
/*
* Licensed to Metamarkets Group Inc. (Metamarkets) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. Metamarkets 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 io.druid.segment.data;
import com.google.common.base.Preconditions;
import com.google.common.io.Closeables;
import com.google.common.primitives.Ints;
import com.google.common.primitives.Shorts;
import com.metamx.common.IAE;
import com.metamx.common.guava.CloseQuietly;
import io.druid.collections.ResourceHolder;
import io.druid.collections.StupidResourceHolder;
import io.druid.segment.CompressedPools;
import java.io.IOException;
import java.nio.ByteBuffer;
import java.nio.ByteOrder;
import java.nio.IntBuffer;
import java.nio.ShortBuffer;
import java.nio.channels.WritableByteChannel;
import java.util.Iterator;
import java.util.List;
public class CompressedVSizeIntsIndexedSupplier implements WritableSupplier<IndexedInts>
{
public static final byte version = 0x2;
private final int totalSize;
private final int sizePer;
private final int numBytes;
private final int bigEndianShift;
private final int littleEndianMask;
private final GenericIndexed<ResourceHolder<ByteBuffer>> baseBuffers;
private final CompressedObjectStrategy.CompressionStrategy compression;
CompressedVSizeIntsIndexedSupplier(
int totalSize,
int sizePer,
int numBytes,
GenericIndexed<ResourceHolder<ByteBuffer>> baseBuffers,
CompressedObjectStrategy.CompressionStrategy compression
)
{
Preconditions.checkArgument(
sizePer == (1 << Integer.numberOfTrailingZeros(sizePer)),
"Number of entries per chunk must be a power of 2"
);
this.totalSize = totalSize;
this.sizePer = sizePer;
this.baseBuffers = baseBuffers;
this.compression = compression;
this.numBytes = numBytes;
this.bigEndianShift = Integer.SIZE - (numBytes << 3); // numBytes * 8
this.littleEndianMask = (int)((1L << (numBytes << 3)) - 1); // set numBytes * 8 lower bits to 1
}
public static int maxIntsInBufferForBytes(int numBytes)
{
int maxSizePer = (CompressedPools.BUFFER_SIZE - bufferPadding(numBytes)) / numBytes;
// round down to the nearest power of 2
return 1 << (Integer.SIZE - 1 - Integer.numberOfLeadingZeros(maxSizePer));
}
private static int bufferPadding(int numBytes)
{
// when numBytes == 3 we need to pad the buffer to allow reading an extra byte
// beyond the end of the last value, since we use buffer.getInt() to read values.
// for numBytes 1, 2 we remove the need for padding by reading bytes or shorts directly.
switch (numBytes) {
case Shorts.BYTES:
case 1:
return 0;
default:
return Ints.BYTES - numBytes;
}
}
public static int maxIntsInBufferForValue(int maxValue)
{
return maxIntsInBufferForBytes(VSizeIndexedInts.getNumBytesForMax(maxValue));
}
public int size()
{
return totalSize;
}
@Override
public IndexedInts get()
{
// optimized versions for int, short, and byte columns
if(numBytes == Ints.BYTES) {
return new CompressedFullSizeIndexedInts();
} else if (numBytes == Shorts.BYTES) {
return new CompressedShortSizeIndexedInts();
} else if (numBytes == 1) {
return new CompressedByteSizeIndexedInts();
} else {
// default version of everything else, i.e. 3-bytes per value
return new CompressedVSizeIndexedInts();
}
}
public long getSerializedSize()
{
return 1 + // version
1 + // numBytes
Ints.BYTES + // totalSize
Ints.BYTES + // sizePer
1 + // compression id
baseBuffers.getSerializedSize(); // data
}
public void writeToChannel(WritableByteChannel channel) throws IOException
{
channel.write(ByteBuffer.wrap(new byte[]{version, (byte) numBytes}));
channel.write(ByteBuffer.wrap(Ints.toByteArray(totalSize)));
channel.write(ByteBuffer.wrap(Ints.toByteArray(sizePer)));
channel.write(ByteBuffer.wrap(new byte[]{compression.getId()}));
baseBuffers.writeToChannel(channel);
}
/**
* For testing. Do not use unless you like things breaking
*/
GenericIndexed<ResourceHolder<ByteBuffer>> getBaseBuffers()
{
return baseBuffers;
}
public static CompressedVSizeIntsIndexedSupplier fromByteBuffer(ByteBuffer buffer, ByteOrder order)
{
byte versionFromBuffer = buffer.get();
if (versionFromBuffer == version) {
final int numBytes = buffer.get();
final int totalSize = buffer.getInt();
final int sizePer = buffer.getInt();
final int chunkBytes = sizePer * numBytes + bufferPadding(numBytes);
final CompressedObjectStrategy.CompressionStrategy compression = CompressedObjectStrategy.CompressionStrategy.forId(buffer.get());
return new CompressedVSizeIntsIndexedSupplier(
totalSize,
sizePer,
numBytes,
GenericIndexed.read(
buffer,
CompressedByteBufferObjectStrategy.getBufferForOrder(order, compression, chunkBytes)
),
compression
);
}
throw new IAE("Unknown version[%s]", versionFromBuffer);
}
public static CompressedVSizeIntsIndexedSupplier fromList(
final List<Integer> list, final int maxValue, final int chunkFactor, final ByteOrder byteOrder, CompressedObjectStrategy.CompressionStrategy compression
)
{
final int numBytes = VSizeIndexedInts.getNumBytesForMax(maxValue);
final int chunkBytes = chunkFactor * numBytes + bufferPadding(numBytes);
Preconditions.checkArgument(
chunkFactor <= maxIntsInBufferForBytes(numBytes),
"Chunks must be <= 64k bytes. chunkFactor was[%s]",
chunkFactor
);
return new CompressedVSizeIntsIndexedSupplier(
list.size(),
chunkFactor,
numBytes,
GenericIndexed.fromIterable(
new Iterable<ResourceHolder<ByteBuffer>>()
{
@Override
public Iterator<ResourceHolder<ByteBuffer>> iterator()
{
return new Iterator<ResourceHolder<ByteBuffer>>()
{
int position = 0;
@Override
public boolean hasNext()
{
return position < list.size();
}
@Override
public ResourceHolder<ByteBuffer> next()
{
ByteBuffer retVal = ByteBuffer
.allocate(chunkBytes)
.order(byteOrder);
if (chunkFactor > list.size() - position) {
retVal.limit((list.size() - position) * numBytes);
} else {
retVal.limit(chunkFactor * numBytes);
}
final List<Integer> ints = list.subList(position, position + retVal.remaining() / numBytes);
final ByteBuffer buf = ByteBuffer
.allocate(Ints.BYTES)
.order(byteOrder);
final boolean bigEndian = byteOrder.equals(ByteOrder.BIG_ENDIAN);
for (int value : ints) {
buf.putInt(0, value);
if (bigEndian) {
retVal.put(buf.array(), Ints.BYTES - numBytes, numBytes);
} else {
retVal.put(buf.array(), 0, numBytes);
}
}
retVal.rewind();
position += retVal.remaining() / numBytes;
return StupidResourceHolder.create(retVal);
}
@Override
public void remove()
{
throw new UnsupportedOperationException();
}
};
}
},
CompressedByteBufferObjectStrategy.getBufferForOrder(byteOrder, compression, chunkBytes)
),
compression
);
}
private class CompressedFullSizeIndexedInts extends CompressedVSizeIndexedInts {
IntBuffer intBuffer;
@Override
protected void loadBuffer(int bufferNum)
{
super.loadBuffer(bufferNum);
intBuffer = buffer.asIntBuffer();
}
@Override
protected int _get(int index)
{
return intBuffer.get(intBuffer.position() + index);
}
}
private class CompressedShortSizeIndexedInts extends CompressedVSizeIndexedInts {
ShortBuffer shortBuffer;
@Override
protected void loadBuffer(int bufferNum)
{
super.loadBuffer(bufferNum);
shortBuffer = buffer.asShortBuffer();
}
@Override
protected int _get(int index)
{
// removes the need for padding
return shortBuffer.get(shortBuffer.position() + index) & 0xFFFF;
}
}
private class CompressedByteSizeIndexedInts extends CompressedVSizeIndexedInts {
@Override
protected int _get(int index)
{
// removes the need for padding
return buffer.get(buffer.position() + index) & 0xFF;
}
}
private class CompressedVSizeIndexedInts implements IndexedInts
{
final Indexed<ResourceHolder<ByteBuffer>> singleThreadedBuffers = baseBuffers.singleThreaded();
final int div = Integer.numberOfTrailingZeros(sizePer);
final int rem = sizePer - 1;
int currIndex = -1;
ResourceHolder<ByteBuffer> holder;
ByteBuffer buffer;
boolean bigEndian;
@Override
public int size()
{
return totalSize;
}
/**
* Returns the value at the given index into the column.
*
* Assumes the number of entries in each decompression buffers is a power of two.
*
* @param index index of the value in the column
* @return the value at the given index
*/
@Override
public int get(int index)
{
// assumes the number of entries in each buffer is a power of 2
final int bufferNum = index >> div;
if (bufferNum != currIndex) {
loadBuffer(bufferNum);
}
return _get(index & rem);
}
/**
* Returns the value at the given index in the current decompression buffer
*
* @param index index of the value in the curent buffer
* @return the value at the given index
*/
protected int _get(final int index)
{
final int pos = buffer.position() + index * numBytes;
// example for numBytes = 3
// big-endian: 0x000c0b0a stored 0c 0b 0a XX, read 0x0c0b0aXX >>> 8
// little-endian: 0x000c0b0a stored 0a 0b 0c XX, read 0xXX0c0b0a & 0x00FFFFFF
return bigEndian ?
buffer.getInt(pos) >>> bigEndianShift :
buffer.getInt(pos) & littleEndianMask;
}
@Override
public Iterator<Integer> iterator()
{
return new IndexedIntsIterator(this);
}
@Override
public void fill(int index, int[] toFill)
{
throw new UnsupportedOperationException("fill not supported");
}
protected void loadBuffer(int bufferNum)
{
CloseQuietly.close(holder);
holder = singleThreadedBuffers.get(bufferNum);
buffer = holder.get();
currIndex = bufferNum;
bigEndian = buffer.order().equals(ByteOrder.BIG_ENDIAN);
}
@Override
public String toString()
{
return "CompressedVSizedIntsIndexedSupplier{" +
"currIndex=" + currIndex +
", sizePer=" + sizePer +
", numChunks=" + singleThreadedBuffers.size() +
", totalSize=" + totalSize +
'}';
}
@Override
public void close() throws IOException
{
Closeables.close(holder, false);
}
}
}

View File

@ -0,0 +1,371 @@
/*
* Licensed to Metamarkets Group Inc. (Metamarkets) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. Metamarkets 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 io.druid.segment.data;
import com.google.common.base.Function;
import com.google.common.collect.Iterables;
import com.google.common.collect.Sets;
import com.google.common.primitives.Ints;
import com.google.common.primitives.Longs;
import com.metamx.common.guava.CloseQuietly;
import io.druid.segment.CompressedPools;
import org.junit.After;
import org.junit.Assert;
import org.junit.Before;
import org.junit.Test;
import org.junit.runner.RunWith;
import org.junit.runners.Parameterized;
import java.io.ByteArrayOutputStream;
import java.io.IOException;
import java.nio.ByteBuffer;
import java.nio.ByteOrder;
import java.nio.channels.Channels;
import java.util.Arrays;
import java.util.Collections;
import java.util.List;
import java.util.Random;
import java.util.Set;
import java.util.concurrent.CountDownLatch;
import java.util.concurrent.atomic.AtomicBoolean;
import java.util.concurrent.atomic.AtomicReference;
@RunWith(Parameterized.class)
public class CompressedVSizeIntsIndexedSupplierTest extends CompressionStrategyTest
{
@Parameterized.Parameters(name = "{index}: compression={0}, byteOrder={1}")
public static Iterable<Object[]> compressionStrategies()
{
final Iterable<CompressedObjectStrategy.CompressionStrategy> compressionStrategies = Iterables.transform(
CompressionStrategyTest.compressionStrategies(),
new Function<Object[], CompressedObjectStrategy.CompressionStrategy>()
{
@Override
public CompressedObjectStrategy.CompressionStrategy apply(Object[] input)
{
return (CompressedObjectStrategy.CompressionStrategy) input[0];
}
}
);
Set<List<Object>> combinations = Sets.cartesianProduct(
Sets.newHashSet(compressionStrategies),
Sets.newHashSet(ByteOrder.BIG_ENDIAN, ByteOrder.LITTLE_ENDIAN)
);
return Iterables.transform(
combinations, new Function<List, Object[]>()
{
@Override
public Object[] apply(List input)
{
return new Object[]{input.get(0), input.get(1)};
}
}
);
}
private static final int[] MAX_VALUES = new int[] { 0xFF, 0xFFFF, 0xFFFFFF, 0x0FFFFFFF };
public CompressedVSizeIntsIndexedSupplierTest(CompressedObjectStrategy.CompressionStrategy compressionStrategy, ByteOrder byteOrder)
{
super(compressionStrategy);
this.byteOrder = byteOrder;
}
private IndexedInts indexed;
private CompressedVSizeIntsIndexedSupplier supplier;
private int[] vals;
private final ByteOrder byteOrder;
@Before
public void setUp() throws Exception
{
CloseQuietly.close(indexed);
indexed = null;
supplier = null;
vals = null;
}
@After
public void tearDown() throws Exception
{
CloseQuietly.close(indexed);
}
private void setupSimple(final int chunkSize)
{
CloseQuietly.close(indexed);
vals = new int[]{0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 16};
supplier = CompressedVSizeIntsIndexedSupplier.fromList(
Ints.asList(vals),
Ints.max(vals),
chunkSize,
ByteOrder.nativeOrder(),
compressionStrategy
);
indexed = supplier.get();
}
private void setupSimpleWithSerde(final int chunkSize) throws IOException
{
vals = new int[]{0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 16};
makeWithSerde(chunkSize);
}
private void makeWithSerde(final int chunkSize) throws IOException
{
CloseQuietly.close(indexed);
ByteArrayOutputStream baos = new ByteArrayOutputStream();
final CompressedVSizeIntsIndexedSupplier theSupplier = CompressedVSizeIntsIndexedSupplier.fromList(
Ints.asList(vals), Ints.max(vals), chunkSize, byteOrder, compressionStrategy
);
theSupplier.writeToChannel(Channels.newChannel(baos));
final byte[] bytes = baos.toByteArray();
Assert.assertEquals(theSupplier.getSerializedSize(), bytes.length);
supplier = CompressedVSizeIntsIndexedSupplier.fromByteBuffer(ByteBuffer.wrap(bytes), byteOrder);
indexed = supplier.get();
}
private void setupLargeChunks(final int chunkSize, final int totalSize, final int maxValue) throws IOException
{
vals = new int[totalSize];
Random rand = new Random(0);
for(int i = 0; i < vals.length; ++i) {
// VSizeIndexed only allows positive values
vals[i] = rand.nextInt(maxValue);
}
makeWithSerde(chunkSize);
}
@Test
public void testSanity() throws Exception
{
setupSimple(2);
Assert.assertEquals(8, supplier.getBaseBuffers().size());
assertIndexMatchesVals();
setupSimple(4);
Assert.assertEquals(4, supplier.getBaseBuffers().size());
assertIndexMatchesVals();
setupSimple(32);
Assert.assertEquals(1, supplier.getBaseBuffers().size());
assertIndexMatchesVals();
}
@Test
public void testLargeChunks() throws Exception
{
for (int maxValue : MAX_VALUES) {
final int maxChunkSize = CompressedVSizeIntsIndexedSupplier.maxIntsInBufferForValue(maxValue);
setupLargeChunks(maxChunkSize, 10 * maxChunkSize, maxValue);
Assert.assertEquals(10, supplier.getBaseBuffers().size());
assertIndexMatchesVals();
setupLargeChunks(maxChunkSize, 10 * maxChunkSize + 1, maxValue);
Assert.assertEquals(11, supplier.getBaseBuffers().size());
assertIndexMatchesVals();
setupLargeChunks(1, 0xFFFF, maxValue);
Assert.assertEquals(0xFFFF, supplier.getBaseBuffers().size());
assertIndexMatchesVals();
setupLargeChunks(maxChunkSize / 2, 10 * (maxChunkSize / 2) + 1, maxValue);
Assert.assertEquals(11, supplier.getBaseBuffers().size());
assertIndexMatchesVals();
}
}
@Test
public void testChunkTooBig() throws Exception
{
for(int maxValue : MAX_VALUES) {
final int maxChunkSize = CompressedVSizeIntsIndexedSupplier.maxIntsInBufferForValue(maxValue);
try {
setupLargeChunks(maxChunkSize + 1, 10 * (maxChunkSize + 1), maxValue);
Assert.fail();
} catch(IllegalArgumentException e) {
Assert.assertTrue("chunk too big for maxValue " + maxValue, true);
}
}
}
@Test
public void testmaxIntsInBuffer() throws Exception
{
Assert.assertEquals(CompressedPools.BUFFER_SIZE, CompressedVSizeIntsIndexedSupplier.maxIntsInBufferForBytes(1));
Assert.assertEquals(CompressedPools.BUFFER_SIZE / 2, CompressedVSizeIntsIndexedSupplier.maxIntsInBufferForBytes(2));
Assert.assertEquals(CompressedPools.BUFFER_SIZE / 4, CompressedVSizeIntsIndexedSupplier.maxIntsInBufferForBytes(4));
Assert.assertEquals(CompressedPools.BUFFER_SIZE, 0x10000); // nearest power of 2 is 2^14
Assert.assertEquals(1 << 14, CompressedVSizeIntsIndexedSupplier.maxIntsInBufferForBytes(3));
}
@Test
public void testSanityWithSerde() throws Exception
{
setupSimpleWithSerde(4);
Assert.assertEquals(4, supplier.getBaseBuffers().size());
assertIndexMatchesVals();
setupSimpleWithSerde(2);
Assert.assertEquals(8, supplier.getBaseBuffers().size());
assertIndexMatchesVals();
}
// This test attempts to cause a race condition with the DirectByteBuffers, it's non-deterministic in causing it,
// which sucks but I can't think of a way to deterministically cause it...
@Test
public void testConcurrentThreadReads() throws Exception
{
setupSimple(4);
final AtomicReference<String> reason = new AtomicReference<>("none");
final int numRuns = 1000;
final CountDownLatch startLatch = new CountDownLatch(1);
final CountDownLatch stopLatch = new CountDownLatch(2);
final AtomicBoolean failureHappened = new AtomicBoolean(false);
new Thread(new Runnable()
{
@Override
public void run()
{
try {
startLatch.await();
}
catch (InterruptedException e) {
failureHappened.set(true);
reason.set("interrupt.");
stopLatch.countDown();
return;
}
try {
for (int i = 0; i < numRuns; ++i) {
for (int j = 0; j < indexed.size(); ++j) {
final long val = vals[j];
final long indexedVal = indexed.get(j);
if (Longs.compare(val, indexedVal) != 0) {
failureHappened.set(true);
reason.set(String.format("Thread1[%d]: %d != %d", j, val, indexedVal));
stopLatch.countDown();
return;
}
}
}
}
catch (Exception e) {
e.printStackTrace();
failureHappened.set(true);
reason.set(e.getMessage());
}
stopLatch.countDown();
}
}).start();
final IndexedInts indexed2 = supplier.get();
try {
new Thread(new Runnable()
{
@Override
public void run()
{
try {
startLatch.await();
}
catch (InterruptedException e) {
stopLatch.countDown();
return;
}
try {
for (int i = 0; i < numRuns; ++i) {
for (int j = indexed2.size() - 1; j >= 0; --j) {
final long val = vals[j];
final long indexedVal = indexed2.get(j);
if (Longs.compare(val, indexedVal) != 0) {
failureHappened.set(true);
reason.set(String.format("Thread2[%d]: %d != %d", j, val, indexedVal));
stopLatch.countDown();
return;
}
}
}
}
catch (Exception e) {
e.printStackTrace();
reason.set(e.getMessage());
failureHappened.set(true);
}
stopLatch.countDown();
}
}).start();
startLatch.countDown();
stopLatch.await();
}
finally {
CloseQuietly.close(indexed2);
}
if (failureHappened.get()) {
Assert.fail("Failure happened. Reason: " + reason.get());
}
}
private void assertIndexMatchesVals()
{
Assert.assertEquals(vals.length, indexed.size());
// sequential access
int[] indices = new int[vals.length];
for (int i = 0; i < indexed.size(); ++i) {
final int expected = vals[i];
final int actual = indexed.get(i);
Assert.assertEquals(expected, actual);
indices[i] = i;
}
Collections.shuffle(Arrays.asList(indices));
// random access
for (int i = 0; i < indexed.size(); ++i) {
int k = indices[i];
Assert.assertEquals(vals[k], indexed.get(k));
}
}
}