org.apache.beam.sdk.io.hbase

Class HBaseIO

java.lang.Object

org.apache.beam.sdk.io.hbase.HBaseIO

public class HBaseIO
extends java.lang.Object

A bounded source and sink for HBase.

For more information, see the online documentation at HBase.

Reading from HBase

The HBase source returns a set of rows from a single table, returning a PCollection<Result>.

To configure a HBase source, you must supply a table id and a Configuration to identify the HBase instance. By default, HBaseIO.Read will read all rows in the table. The row range to be read can optionally be restricted using with a Scan object or using the HBaseIO.Read.withKeyRange(org.apache.beam.sdk.io.range.ByteKeyRange), and a Filter using HBaseIO.Read.withFilter(org.apache.hadoop.hbase.filter.Filter), for example:

 // Scan the entire table.
 p.apply("read",
     HBaseIO.read()
         .withConfiguration(configuration)
         .withTableId("table"));

 // Filter data using a HBaseIO Scan
 Scan scan = ...
 p.apply("read",
     HBaseIO.read()
         .withConfiguration(configuration)
         .withTableId("table"))
         .withScan(scan));

 // Scan a prefix of the table.
 ByteKeyRange keyRange = ...;
 p.apply("read",
     HBaseIO.read()
         .withConfiguration(configuration)
         .withTableId("table")
         .withKeyRange(keyRange));

 // Scan a subset of rows that match the specified row filter.
 p.apply("filtered read",
     HBaseIO.read()
         .withConfiguration(configuration)
         .withTableId("table")
         .withFilter(filter));
 

readAll() allows to execute multiple Scans to multiple Tables. These queries are encapsulated via an initial PCollection of HBaseIO.Reads and can be used to create advanced compositional patterns like reading from a Source and then based on the data create new HBase scans.

Note: HBaseIO.ReadAll only works with runners that support Splittable DoFn.

 PCollection<Read> queries = ...;
 queries.apply("readAll", HBaseIO.readAll().withConfiguration(configuration));
 

Writing to HBase

Writing Mutation

The HBase sink executes a set of row mutations on a single table. It takes as input a PCollection<Mutation>, where each Mutation represents an idempotent transformation on a row.

To configure a HBase sink, you must supply a table id and a Configuration to identify the HBase instance, for example:

 Configuration configuration = ...;
 PCollection<Mutation> data = ...;

 data.apply("write",
     HBaseIO.write()
         .withConfiguration(configuration)
         .withTableId("table"));
 

Writing RowMutations

An alternative way to write to HBase is with writeRowMutations(), which takes as input a PCollection>, representing KVs of bytes row keys and RowMutations.

This implementation is useful for preserving mutation order if the upstream is ordered by row key, as RowMutations will only be applied after previous RowMutations are successful.

To configure the sink, you must supply a table id string and a Configuration to identify the HBase instance, for example:

 Configuration configuration = ...;
 PCollection<KV<byte[], RowMutations>> data = ...;

 data.apply("write",
     HBaseIO.writeRowMutations()
         .withConfiguration(configuration)
         .withTableId("table"));
 

Note that the transformation emits the number of RowMutations written as an integer after successfully writing to HBase.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	HBaseIO.Read A PTransform that reads from HBase.
static class	HBaseIO.ReadAll Implementation of readAll().
static class	HBaseIO.Write A PTransform that writes to HBase.
static class	HBaseIO.WriteRowMutations Transformation that writes RowMutation objects to a Hbase table.

Method Summary

Modifier and Type	Method and Description
static HBaseIO.Read	read() Creates an uninitialized HBaseIO.Read.
static HBaseIO.ReadAll	readAll() A PTransform that works like read(), but executes read operations coming from a PCollection of HBaseIO.Read.
static HBaseIO.Write	write() Creates an uninitialized HBaseIO.Write.
static HBaseIO.WriteRowMutations	writeRowMutations()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Detail

read

public static HBaseIO.Read read()

Creates an uninitialized HBaseIO.Read. Before use, the Read must be initialized with a HBaseIO.Read.withConfiguration(Configuration) that specifies the HBase instance, and a tableId that specifies which table to read. A Filter may also optionally be specified using HBaseIO.Read.withFilter(org.apache.hadoop.hbase.filter.Filter).

readAll

public static HBaseIO.ReadAll readAll()

A PTransform that works like read(), but executes read operations coming from a PCollection of HBaseIO.Read.

write

public static HBaseIO.Write write()

Creates an uninitialized HBaseIO.Write. Before use, the Write must be initialized with a HBaseIO.Write.withConfiguration(Configuration) that specifies the destination HBase instance, and a tableId that specifies which table to write.

writeRowMutations

public static HBaseIO.WriteRowMutations writeRowMutations()