Oracle Technology Supports CERN's New Particle Accelerator Experiments
CERN, the world's largest particle physics laboratory, is preparing for the full deployment of the Large Hadron Collider in May next year. The experiments to be conducted by this new particle accelerator will make unprecedented computational demands on CERN's database systems. Collaboration with Oracle is allowing the laboratory to address these issues.
The Large Hadron Collider (LHC) is the world's largest particle accelerator and collider. Currently under construction near Geneva by CERN, the collider is on target to be officially deployed in May 2008. The LHC is the result of funding and collaboration between over 10,000 scientists and engineers from around 500 academic institutes, laboratories, educational institutions and industrial companies worldwide. Equipment is being built in many European countries, as well as in Canada, India, Japan, Russia and the US.
In order to process the computationally-intensive data that the LHC will generate, CERN has established the LHC Computing Grid (LCG). The data stored in the Grid's databases is crucial for the operation of LCG services as well as for processing the data generated by experiments. In addition, a large amount of non-event data, such as detector conditions, calibration and production book-keeping, is stored within these databases. Because these relational databases are so important to the success of the LHC experiments, it is vital that they are supremely reliable, scalable and available. To ensure these conditions, CERN has been working closely with Oracle on the Distributed Deployment of Databases Project (3D).
The 3D project is a collaboration between CERN IT department's Physics Services Support (IT/PSS) group, the LHC experiments and the LCG sites. With help from Oracle, the project ensures that the databases offer crucial services for the LHC experiments. These services include ensuring that there is consistent and available storage for data that is simultaneously accessed or updated; the potential for recovery to a consistent state in the case of hardware, software or human error; and support for ad-hoc queries.
The project is using an Oracle Database cluster to ensure reliable database services at different stages of the data flow. Oracle's Real Application Clusters (RAC) technology allows CERN to deploy a single database across a cluster of servers. Each cluster consists of multiple server nodes, which access shared data. As the LHC experiments begin, data volume and access requirements may change. The scalability of Oracle RAC clusters will allow CERN to add, remove or reassign resources as appropriate. In addition, if one server node experiences damage or failure, another can automatically pick up the workload, ensuring the database is still accessible while the hardware is repaired.
"Oracle Real Application Clusters and Oracle Streams Replication are the two cornerstones of the distributed database architecture for CERN's Large Hadron Collider project," says Dirk Düllmann, leader of the LCG Distributed Database Deployment project. "The joint work with Oracle as part of the CERN openlab has allowed us to build for the first time a scalable and truly distributed database environment serving the physics community both at CERN and ten major sites worldwide."
The IT/PSS group has worked hard to ensure that the physics database service is ready for the start up of the LHC. It is now live on Oracle Database 10g running on clusters of Linux servers. More than one hundred database server nodes are in operation across fifteen clusters, serving almost two million database sessions a week. The success of this new architecture at CERN has led to the introduction of similar installations at partner sites worldwide, forming one of the largest Oracle RAC installations in the world.
With the full deployment of the Large Hadron Collider in May 2008, CERN will again be leading the way in scientific research and experimentation.
