Beneath the SURFace: An MRI-like View into the Life of a 21st-Century Datacenter

Real-world data is crucial in understanding and improving our world, from health care to datacenters. To help the computer systems community with data-driven decisions, we open-source a collection of fine-grained, low-level operational logs from the largest public-sector datacenter in the Netherlands (SURFsara). In this article, we describe the infrastructure providing the data, give examples of some of this data, and perform thorough statistical analysis to indicate that this ongoing collection not only reflects the ground truth but will be useful to designers and maintainers of large clusters, and generally to computer systems practitioners.