I'm interested in using Avro to save and read files from Hadoop HDFS and I saw some Jira's in Hadoop issue tracker regarding implementing support for Avro but there were no examples how to enable Avro support in Hadoop. Also I'm not completely sure that current 0.20 has support for Avro because some Jira's were closed for 0.21. Is it possible to get latest 0.21 and end somehow turn on Avro and use it?
A:
I am using Avro with Hadoop 0.20 (Amazon Elastic MapReduce). You enable Avro support by including the Avro jar in your classpath, then extending Avro provided classes to implement your Map and Reduce tasks.
See http://avro.apache.org/docs/1.4.0/api/java/index.html and particularly the mapred package at http://avro.apache.org/docs/1.4.0/api/java/org/apache/avro/mapred/package-frame.html .
Spike Gronim
2010-09-24 00:33:14