Architecture

HareDB, the total solution we provided, is a hybrid distributed system including Hadoop platform, HBase, and related data management tools. With the HareDB solution, users will find some easier way to get involved with the big data. The layers of HareDB are shown in the following figure.

HareDB is based on the Hadoop system with the HDFS, MapReduce, and Hive. As HBase resides in the Hadoop ecosystem as the NoSQL database, our enhancement of the HBase provides the SQL-like support for the users who use the RDB in common. With HareDB, you may reduce a lot of efforts to re-build your application. It is one of the finest solutions to store, compute, and do something on your tedious big data.

HareDB HBase Client is the tool for users to operate the HBase just as the actions they did in the RDB. Friendly user interface enables the users to have the easiest way to access the HBase. Since SQL-like command helps the user to execute the query in a common way, HareDB HBase Client provides the friendly user interface which allows the user to use SQL-command and Hive/Pig script to access the data and see the result in a faster and clear way.

HareDB Core implements the engine of accessing the data in HBase via HBase API and Coprocessor. Instead of using Hive to access the HDFS layer and produce the MapReduce jobs to execute the query sent by the user, we accept the SQL from users and execute the query through HBase API and Coprocessors.

For those users who wants to apply our HareDB core to their application which is customized by themselves, the RESTful service can provide brilliant API for people who wants to directly access the HBase. HareDB ODBC/JDBC drivers also support the customers on the integration purpose. These features expands the usage of the HBase and brings the brilliant methodology for HBase users and developers to export the data in HBase. With the HareDB Thrift server stands in back of our HareDB core, HareDB provides abundant ways to increase the accessibility.