HBase Compound Indexes

HBase Compound Indexes

We recently set up HBase and HBase-trx (from https://github.com/hbase-trx) to use multiple-column indexes with this code.  After you compile it, just copy the jar and the hbase-trx jar into your hbase’s lib folder and you should be good to to!

When you create a composite index, you can see the metadata for the index by looking at the table description.  One of the properties will read “INDEXES =>” followed by index names and ‘family:qualifier’ style column names in the index.

KeyGeneratorFactory:

package com.ir.store.hbase.indexes;

import java.util.List;

import org.apache.hadoop.hbase.client.tableindexed.IndexKeyGenerator;

public class KeyGeneratorFactory {

public static IndexKeyGenerator getInstance(List columns) {
return new HBaseIndexKeyGenerator(columns);
}
}

HBaseIndexKeyGenerator:

package com.ir.store.hbase.indexes;

import java.io.DataInput;
import java.io.DataOutput;
import java.io.IOException;
import java.util.ArrayList;
import java.util.List;
import java.util.Map;
import org.apache.hadoop.hbase.client.tableindexed.IndexKeyGenerator;
import org.apache.hadoop.hbase.util.Bytes;

public class HBaseIndexKeyGenerator extends Object implements IndexKeyGenerator {
public static byte[] KEYSEPERATOR = "~;?".getBytes();

private int columnCount;
private List columnNames = new ArrayList();

public HBaseIndexKeyGenerator(List memberColumns) {
// For new key generators
columnNames = memberColumns;
columnCount = memberColumns.size();
}

public HBaseIndexKeyGenerator() {
// Hollow constructor for deserializing -- should call readFields shortly
columnCount = 0;
}

public void readFields(DataInput binaryInput) throws IOException {
columnCount = binaryInput.readInt();
for (int currentColumn = 0; currentColumn < columnCount; currentColumn++)
columnNames.add(Bytes.readByteArray(binaryInput));
}

public void write(DataOutput binaryOutput) throws IOException {
binaryOutput.writeInt(columnCount);
for (byte[] columnName : columnNames)
Bytes.writeByteArray(binaryOutput, columnName);
}

public byte[] createIndexKey(byte[] baseRowIdentifier, Map baseRowData) {
byte[] indexRowIdentifier = null;
for (byte[] columnName: columnNames) {
if (indexRowIdentifier == null)
indexRowIdentifier = baseRowData.get(columnName);
else indexRowIdentifier = Bytes.add(indexRowIdentifier, HBaseIndexKeyGenerator.KEYSEPERATOR, baseRowData.get(columnName));
}
if (baseRowIdentifier != null)
return Bytes.add(indexRowIdentifier, HBaseIndexKeyGenerator.KEYSEPERATOR, baseRowIdentifier);
return indexRowIdentifier;
}
}