type: languages title: “Beam Java SDK Extensions”

Apache Beam Java SDK Extensions

Join-library

Join-library provides inner join, outer left join, and outer right join functions. The aim is to simplify the most common cases of join to a simple function call.

The functions are generic and support joins of any Beam-supported types. Input to the join functions are PCollections of Key / Values. Both the left and right PCollections need the same type for the key. All the join functions return a Key / Value where Key is the join key and value is a Key / Value where the key is the left value and right is the value.

For outer joins, the user must provide a value that represents null because null cannot be serialized.

Example usage:

PCollection<KV<String, String>> leftPcollection = ...
PCollection<KV<String, Long>> rightPcollection = ...

PCollection<KV<String, KV<String, Long>>> joinedPcollection =
  Join.innerJoin(leftPcollection, rightPcollection);

Sorter

This module provides the SortValues transform, which takes a PCollection<KV<K, Iterable<KV<K2, V>>>> and produces a PCollection<KV<K, Iterable<KV<K2, V>>>> where, for each primary key K the paired Iterable<KV<K2, V>> has been sorted by the byte encoding of secondary key (K2). It is an efficient and scalable sorter for iterables, even if they are large (do not fit in memory).

Caveats

  • This transform performs value-only sorting; the iterable accompanying each key is sorted, but there is no relationship between different keys, as Beam does not support any defined relationship between different elements in a PCollection.
  • Each Iterable<KV<K2, V>> is sorted on a single worker using local memory and disk. This means that SortValues may be a performance and/or scalability bottleneck when used in different pipelines. For example, users are discouraged from using SortValues on a PCollection of a single element to globally sort a large PCollection. A (rough) estimate of the number of bytes of disk space utilized if sorting spills to disk is numRecords * (numSecondaryKeyBytesPerRecord + numValueBytesPerRecord + 16) * 3.

Options

  • The user can customize the temporary location used if sorting requires spilling to disk and the maximum amount of memory to use by creating a custom instance of BufferedExternalSorter.Options to pass into SortValues.create.

Example usage of SortValues

PCollection<KV<String, KV<String, Integer>>> input = ...

// Group by primary key, bringing <SecondaryKey, Value> pairs for the same key together.
PCollection<KV<String, Iterable<KV<String, Integer>>>> grouped =
    input.apply(GroupByKey.<String, KV<String, Integer>>create());

// For every primary key, sort the iterable of <SecondaryKey, Value> pairs by secondary key.
PCollection<KV<String, Iterable<KV<String, Integer>>>> groupedAndSorted =
    grouped.apply(
        SortValues.<String, String, Integer>create(BufferedExternalSorter.options()));