Speaker
Description
The HEP experiments and community face complex computing and storage requirements which are expected to increase by several factors with the advent of HL-LHC. By the end of Run3, the machine will have accumulated roughly 10% of the total dataset. These data, stored in the ROOT data format, are then analysed by the various experiment communities, with varying levels of coordination and centralisation.
Very often, the computational complexity of these physics analyses is such that parallelisation must be employed to ensure a reasonable time-to-insight for the physicists. Batch computing on the grid is a well-oiled workflow that has enabled obtaining a plethora of important physics results. In recent times, interactive scheduling approaches have emerged that lower the programming entry barrier and allow for a more ergonomic end-user experience. In this context, a specific combination of storage, computing resources, software distribution and distributed engines can be identified and is sometimes referred to as an "analysis facility".
In this contribution, we highlight existing technologies such as ROOT, SWAN, EOS, XRootD and others that represent concrete building blocks available in production today to help the community in addressing the expected complexity of HL-LHC analyses. We also display how various current and future physics use cases can benefit from such technologies. These are used both in analyses facilities and other distributed computing infrastructures to minimise the time-to-insight and enable ergonomic and user-friendly distributed computing.
