CORAL Server and CORAL Server Proxy: Scalable Access to Relational Databases from CORAL Applications

The CORAL software is widely used at CERN for accessing the data stored by the LHC experiments using relational database technologies. CORAL provides a C++ abstraction layer that supports data persistency for several backends and deployment models, including local access to SQLite files, direct client access to Oracle and MySQL servers, and read-only access to Oracle through the FroNTier web server and cache. Two new components have recently been added to CORAL to implement a model involving a middle tier "CORAL server" deployed close to the database and a tree of "CORAL server proxy" instances, with data caching and multiplexing functionalities, deployed close to the client. The new components are meant to provide advantages for read-only and read-write data access, in both offline and online use cases, in the areas of scalability and performance (multiplexing for several incoming connections, optional data caching) and security (authentication via proxy certificates). A first implementation of the two new components has been released in the summer 2009 and is presently fully deployed in the ATLAS online system, where the CORAL server and proxy are used to read the configuration and conditions data that are needed by the High Level Trigger, allowing the configuration of a farm of several thousand processes. This software is the result of a joint development of the CERN IT Department and of the Mainz and SLAC groups of the ATLA S collaboration. This presentation will report on the status of developments and deployment for the CORAL server and proxy components and on their usage in the ATLAS online system during the first year of LHC data taking. The design of the software architecture and the outlook for its eventual use for offline data acccess will also be covered.