Oracle DataSource for Apache Hadoop (OD4H): introduction



Currently we see that Hadoop is becoming part of Enterprise Data Warehouse family. But family should be connected to each other. Sometimes we need access to Hadoop from Oracle Database. Sometimes Hadoop users need enterprise data stored in Oracle database.

Hive has very interesting concept — External Tables which allow you to define Java classes to access external database and present it as a native hive table.

Oracle Datasource for Apache Hadoop (formerly Oracle Table Access for Apache Hadoop) turns Oracle Database tables into a Hadoop data source (i.e., external table) enabling direct and consistent Hive QL/Spark SQL queries, as well as direct Hadoop API access. Applications can join master data or dimension data in Oracle Database with data stored in Hadoop. Additionally data can be written back to Oracle Database after processing.

Oracle Datasource for Apache Hadoop optimizes a query’s execution plans using predicate and projection pushdown, and partition pruning. Database table access is performed in parallel based on the selected split patterns, using smart and secure connections (Kerberos, SSL, Oracle Wallet), regulated by both Hadoop (i.e., maximum concurrent tasks) and Oracle DBAs (i.e., max pool size).

Continue reading ‘Oracle DataSource for Apache Hadoop (OD4H): introduction’ »