Somerville

The Somerville on-premises cloud platform is designed to support data-intensive use cases from survey astronomy, including very large databases (with trillions of records) and real-time data streaming and analysis. 

Enabling unprecedented research

Named in honour of Scottish mathematician and astronomer Mary Somerville, the Somerville service supports astronomers internationally to pursue their science ambitions. 

Providing research-cloud services for the STFC IRIS Programme, Somerville hosts production science for the Vera C. Rubin Observatory, the European Space Agency’s Gaia satellite, and the UK wide-field astronomy archives. 

At the time of writing, the largest database hosted on Somerville has more than 120 billion records of astronomy observations – that is 15 records for each and every person living on Earth.

Vera C. Rubin Observatory 

The Rubin Observatory is conducting a ten-year survey of the Southern Hemisphere sky (the Legacy Survey of Space and Time, or LSST) using the world’s largest digital camera. Rubin’s science goals include understanding the nature of dark matter and dark energy, creating an inventory of the Solar System, mapping the Milky Way, and exploring objects that change position or brightness over time.

Facilitating data science 

The volume and complexity of data that Rubin is capturing challenges the capabilities of research computing worldwide. Somerville is the host for the UK’s Rubin Independent Data Access Centre (DAC) and, over the course of the LSST, will connect astronomers with at least 200 Petabytes of survey products and provide advanced tooling to handle these products.

EPCC, in collaboration with the Institute for Astronomy at the University of Edinburgh, has a critical role in the design and operation of the UK DAC, which is expected to be in service until at least 2036. 

Technology

Somerville is installed at the University of Edinburgh’s Advanced Computing Facility, and managed by a joint team from EPCC and the University’s Institute of Astronomy. 

The system is built on Scientific OpenStack. This variant of the popular OpenStack toolkit has been customised by StackHPC Limited to meet research-computing capabilities.

Nodes

• 5,760 (virtualized) CPU cores with 10 Terabytes of memory.

Interconnect technologies

• 100 Gbps Ethernet internal data network
• 2×100 Gbps uplink to UK JANET research network

System storage 

• 10 PB storage (Ceph, scientific-data archives)
• 100 TB NVMe storage (low-latency databasing)

Rubin Independent Data Access Centre

The Rubin Independent Data Access Centre (DAC) provides three different user interfaces to LSST: 

• A web-based science portal for interactive enquiries
• A notebook-based engine for more substantial and scripted analysis workflows
• An HPC/batch interface for the most ambitious and computationally intensive research campaigns.

These interfaces are backed by custom services, engineered within the Rubin project to the requirements of the survey. They include a bespoke distributed database system called Qserv, able to serve LSST catalogues containing hundreds of billions of objects and to satisfy the demands of hundreds of concurrent queries.

Funding

Somerville is funded by STFC via the LSST:UK and IRIS programmes.

Further information

Rubin Observatory 
https://rubinobservatory.org

Wide Field Astronomy Unit 
https://www.roe.ac.uk/ifa/wfau/

Gaia 
https://www.esa.int/Science_Exploration/Space_Science/Gaia

LSST:UK 
https://www.lsst.ac.uk

IRIS 
https://www.iris.ac.uk

Access

To access Somerville, please see: 
https://www.iris.ac.uk/rsap