Java Developer Magazine on Ulitzer

The real-time embedded database folks at McObject have slapped a Java Native Interface (JNI) interface on their eXtremeDB widgetry and claim it's the fastest possible database solution in Java-dom.

The move is supposed to underscore their attempt to broaden the product's appeal by pushing it out to developers of Java-based set-top boxes and other embedded systems who get a proven off-the-shelf database with minimal RAM and a small 100KB CPU footprint in return.

eXtremeDB JNI also targets creators of real-time enterprise systems such as highly scalable finance, e-commerce, chi-chi social networking, and other web-based applications where it adds the benefits of an in-memory object cache, the company says, without the hassle of some so-called "NoSQL" solutions, such as high storage demands and lack of safeguards on data integrity.

McObject says eXtremeDB JNI delivers the speed of compiled C/C++ for performance-intensive data sorting, storage and retrieval functions, with the chumminess of a familiar language. "Nobody else is using Java this way," it says.

eXtremeDB manages data entirely in main memory, eliminating file I/O, cache management and the other overhead found in database systems that incorporate disk storage.

McObjects claims eXtremeDB JNI dramatically outperforms the traditional object caching software used to accelerate highly scalable enterprise (non-embedded systems) applications. In this scenario, developers deploy the 64-bit edition of eXtremeDB (eXtremeDB-64) as an in-memory front-end for an enterprise RDBMS.

The JNI enhances eXtremeDB's value as a cache by leveraging Java's familiarity among web and enterprise application developers. It offers the rapid in-memory data access provided by NoSQL distributed object cache solutions such as memCached, but with persistence: in the event of a software or hardware failure, the cached data can be easily recovered through eXtremeDB features such as transaction logging, database replication (eXtremeDB high availability), and the optional disk- or flash-based storage provided by McObject's eXtremeDB Fusion hybrid database technology.

Reportedly web application developers who have switched from memCached to eXtremeDB say that McObject's in-memory database provides greater storage efficiency. Its performance and multiple indexes enable end-user data to be generated on-the-fly with a smaller volume of core data stored in the database.

One of the big social networks is reportedly using eXtremeDB JNI to store an in-memory copy of the connection information used by members to meet other members, find out who's online, look up friends and other functions. A comprehensive dataset is stored permanently in an Oracle DBMS, but by keeping the most frequently used records accessible in memory with the eXtremeDB embedded database, unnecessary hits to the Oracle back end are avoided and the company has reportedly found its application runs much faster.