Package org.apache.trevni.avro.mapreduce

Class AvroTrevniKeyValueOutputFormat<K,V>

java.lang.Object
org.apache.hadoop.mapreduce.OutputFormat<AvroKey<K>,AvroValue<V>>
org.apache.hadoop.mapreduce.lib.output.FileOutputFormat<AvroKey<K>,AvroValue<V>>
org.apache.trevni.avro.mapreduce.AvroTrevniKeyValueOutputFormat<K,V>
Type Parameters:
K - The type of key. If an Avro type, it must be wrapped in an AvroKey.
V - The type of value. If an Avro type, it must be wrapped in an AvroValue.

Writes a directory of files per task, each comprising a single filesystem block. To reduce the number of files, increase the default filesystem block size for the job. Each task also requires enough memory to buffer a filesystem block.

public class AvroTrevniKeyValueOutputFormat<K,V> extends FileOutputFormat<AvroKey<K>,AvroValue<V>>
An OutputFormat that writes Avro data to Trevni files. This implement was modeled off AvroKeyValueOutputFormat to allow for easy transition * FileOutputFormat for writing Trevni container files of key/value pairs.

Since Trevni container files can only contain records (not key/value pairs), this output format puts the key and value into an Avro generic record with two fields, named 'key' and 'value'.

The keys and values given to this output format may be Avro objects wrapped in AvroKey or AvroValue objects. The basic Writable types are also supported (e.g., IntWritable, Text); they will be converted to their corresponding Avro types.