Data Grid (Database in-memory)

With Oracle being the major database and acquired tangosol coherence & BEA . IBM's data grid products on demand (websphere). The need for caching and holding the data in-memory is becoming more and more prominent in performance intensive applications(Ex: Google, Amazon etc.).

Advent of Virtualization and Cloud computing around the corner adds to this. The days of hitting the database from the application would fade off. This means its not very important where your data resides as long as you are able to read & write them efficiently.

This brings in need for standardization and need for interoperability between the applications -> data grids -> databases. Out of curiosity went and found a very old JSR 107 which to me needs revamp with the new progress in the distributed caching market. With microsoft data grid products in the mind we need specification/standardization across platforms.

Key Challenges,
1. Interoperability standards between distributed caching products in the market.
2. Interoperability standards between applications and data grid products.
3. Mix match database and data grid products from different vendors.

Possible outcomes,
1. Databases handling the the data grid completely which means caching the most used data with limited flexibility to change the caching algorithm and what you want to cache.
2. Leverage the ORM capabilities and standardization, and make them data grid aware. Which means very minimal change to the application developers.
3. Update the JDBC API specification to support data grid. Which means data grid/database vendor implement in such a way it abstracts the data grid from application users.


Reference:

IBM Webspehere with datagrid

Oracle

Real time drives database virtualization