The CERN Accelerator Measurement Database: On The Road To Federation

The Measurement database, acting as short-term central persistence and front-end of the CERN accelerator Logging Service, receives billions of time-series data per day for 200,000+ signals. A variety of data acquisition systems on hundreds of front-end computers publish source data that eventually end up being logged in the Measurement database. As part of a federated approach to data management, information about source devices are defined in a Configuration database, whilst the signals to be logged are defined in the Measurement database. A mapping, which is often complex and subject to change/extension, is required in order to subscribe to the source devices, and write the published data to the corresponding named signals. Since 2005, this mapping was done by means of dozens of XML files, which were manually maintained by multiple persons, resulting in a configuration that was error prone. In 2010 this configuration was fully centralized in the Measurement database itself, reducing significantly the complexity and the actors in the process. Furthermore, logging processes immediately pick up modified configurations via JMS based notifications sent directly from the database. This paper will describe the architecture and the benefits of current implementation, as well as the next steps on the road to a fully federated solution.