Distributed Database Computing with GridGain

Distributed MySQL, PostgreSQL, Oracle and RDBMS Databases

The GridGain in-memory computing platform can be installed between the application and data layers and uploads data from the underlying RDBMS, NoSQL or Hadoop datastores from disk into RAM. The results is massive scalability and a 1,000x or greater improvement in processing speed without manually sharding the underlying database. This is accomplished by creating a distributed in-memory data grid across the nodes in the GridGain cluster. The system accelerates compute by moving data from disk into RAM and also by leveraging the GridGain Compute Grid to use all of the compute power of the cluster for massive parallel processing. The system is highly scalable because new nodes can be added to the GridGain cluster at any time and the automatic rebalancing feature will balance the in-memory data between nodes.

When using the GridGain Persistent Store feature, the superset of data and all the SQL indexes are kept on disk, making GridGain fully operational from disk. Users can then define whether all or only a portion of the full dataset is kept in-memory. The combination of Persistent Store and the platform's advanced SQL capabilities allows GridGain to serve as a distributed transactional SQL database, spanning both memory and disk.

GridGain supports ACID transactions so data changes at the application layer are written to the underlying RDBMS. The GridGain In-Memory SQL Grid allows you to interact with the GridGain cluster using standard SQL commands including DML and DDL support. The result is a platform which can power OLTP, OLAP and hybrid transactional/analytical processing (HTAP) use cases.

GridGain removes the scalability limitations of many common relational databases and allows you to create:

GridGain is ANSI SQL-99 compliant with full data indexing so it can support distributed NoSQL and HDFS while also providing a number of database-specific advantages such as: