Widespread distributed processing of big datasets has been around for more than a decade now thanks to Hadoop, but only recently higher-level abstractions have been proposed for programmers to easily operate on those datasets, e.g. Spark. ROOT has joined that trend with its RDataFrame tool for declarative analysis, which currently supports local multi-threaded parallelisation. However, RDataFrame’s programming model is general enough to accommodate multiple implementations or backends: users could write their code once and execute it as-is locally or distributedly, just by selecting the corresponding backend. This abstract introduces PyRDF, a new python library developed on top of RDataFrame to seamlessly switch from local to distributed environments with no changes in the application code. In addition, PyRDF has been integrated with a service for web-based analysis, SWAN, where users can dynamically plug in new resources, as well as write, execute, monitor and debug distributed applications via an intuitive interface.

Thanks to its RDataFrame interface, ROOT now supports the execution of the same physics analysis code both on a single machine and on a cluster of distributed resources. In the latter scenario, it is common to read the input ROOT datasets over the network from remote storage systems, which often increases the time it takes for physicists to obtain their results. Storing the remote files much closer to where the computations will run can bring latency and execution time down. Such a solution can be improved further by caching only the actual portion of the dataset that will be processed on each machine in the cluster, reusing it in subsequent executions on the same input data. This paper shows the benefits of applying different means of caching input data in a distributed ROOT RDataFrame analysis. Two such mechanisms will be applied to this kind of workflow with different configurations, namely caching on the same nodes that process data or caching on a separate server.

No abstract