ASTRONOMY TAKES TO THE CLOUD

[...]the next-generation Very Large Array (ngVLA) will generate hundreds of petabytes annually, roughly 1,000 times more than the VLA does today, says Brian Glendenning, assistant director for data management and software at the National Radio Astronomy Observatory, who is based in Albuquerque, New Mexico Such data sets are out of reach for conventional workflows: it isn't feasible to download that much data and store them locally, says Mario Juric, an astronomer at the University of Washington in Seattle. No problem "There are definitely times when projects involving just mid-size data can see a lot of benefits from cloud computing," says Ivelina Momcheva, a mission scientist at the Space Telescope Science Institute in Baltimore, Maryland. In 2015, when Momcheva and her colleagues had only an 8-core server available for their 3D-HST project, which analysed data from the Hubble Space Telescope to better understand the forces that shape galaxies in the distant Universe. According to estimates from the University of Washington, for instance, cloud-based processes that cost $43 per experiment cost only $6 after a few months of optimization, Juric says. "With STRIDES, if an institution commits a given amount of their dollars on grants, then the Googles and Microsofts and Amazons of the world can compete with each other so investigators can get the best deals with vendors," Bourne says. Since its inception in 2018, STRIDES has helped researchers with more than 225 projects exceeding 20 million total computing hours, and saved roughly $6 million, says Susan Gregurick, who leads the NIH's strategic plan for data science in Bethesda, Maryland.