The CMS PhEDEx System: a Novel Approach to Robust Grid Data Distribution

The CMS experiment has taken a novel approach to Grid data distribution. Instead of having a central processing component making global decisions on replica allocation, CMS has a data management layer composed of a series of collaborating agents; the agents are persistent, stateless processes which manage specific parts of replication operations at each site in the distribution network. The agents communicate asynchronously through a blackboard architecture; as a result the system is robust in the face of many local failures. CMS’ data distribution network is represented in a similar fashion to the internet. This means that the management of multi-hop transfers at dataset level is simple. This system allows CMS to seamlessly bridge the gap between traditional HEP “waterfall” type data distribution, and more dynamic Grid data distribution. Using this system CMS is able to reach transfer rates of 500Mbps, and sustain transfers for long periods in tape-to-tape transfers. The system has been used to manage branches of large scale Service Challenges for LCG this year.