From molecular biology and chemistry to astronomy, earth sciences, and particle physics, modern experimental science relies increasingly on the acquisition, manipulation, and processing of large amounts of data, and the systematic orchestration of computationally intensive simulations and analyses. Much of the analysis is expensive in terms of the data processing and storage resources required, and it sometimes runs for weeks and needs careful monitoring. It is also laboriously repetitive as scientists explore different settings for their algorithms, data sets are continually updated from instruments, and new information prompts the whole “in silico” (computer-simulated) experiment to be rerun and cross-checked.