Today's applications require faster and more frequent access to data at the mid-tier than ever before. This is due to a number of factors, including massive growth of data volumes and the extreme processing requirements that accompany such growth, the pressure from ever-changing business requirements, and the adoption of architectural approaches and frameworks such as Service Oriented Architecture (SOA) and Web 2.0 and the resulting demands that these frameworks make on data. The requirements of accessing and modifying this data quickly using many different forms such as batch, OLTP, and online access are stretching back-end data stores to the limit.

Much of this overload is caused by the classic mismatch between the object and relational worlds and the massive overhead required to convert data between the two different formats. The problem is compounded further when data is accessed and changed at the mid-tier by thousands of users and processes. The only real solution for bringing high-performance transaction processing to Java and other object languages is to store the data in object form in a reliable and scalable mid-tier cache.