In the Scientific Computing Department (SCD) we research, develop and support leading edge high-performance computational & data storage infrastructures and scientific software to perform and support world class science. We have over 240 staff in roles such as systems management, research software engineering, open science and computational science developing projects and services to make a real difference to the scientific communities we support. We help our communities by providing systems and software to help analyse, simulate and visualise research data.
Graduates in SCD typically experience 4 projects within the department and one 3 month project in another department.
Research Infrastructure Group – Systems Tools
The Scientific Computing Department has a comprehensive computing and data infrastructure supporting STFC’s World Class Science. The infrastructure supports 100s PB data, 10,000s compute cores and high speed networking. To ensure effective operation of this large and complex infrastructure support tools are needed such as:
- Systems/service exception monitoring, event handling and alerting
- Performance monitoring and trend analysis for systems and networks
- Centralised logging servers for system and application logs with search functions
- Machine room, rack and systems environmental and power monitoring and trend analysis
- A variety of inventory and configuration management systems covering such items as systems configuration, rack layouts, power connections, cabling, service dependencies and IP address management
- System power control and access to remote management interfaces
- Backup monitoring
You’ll be part of a team responsible for updating and implementing new tools such as:
- Replacement of the existing ganglia systems with data collectors feeding influxDB with Grafana frontends
- Standardise on LibraNMS as a network monitoring system and ensure previous networking tools in use are retired
- Replacement of multiple configuration and inventory systems with theopen source tool NetBox
Dynamic Infrastructure Group – Cloud Team
The STFC Cloud is a service based on OpenStack which delivers compute resources to scientific users both within STFC and externally. This project forms part of efforts to widen use of the STFC Cloud across STFC.
You’ll be responsible for developing a compute service to provide access to the Cloud and other departmental compute resources using Jupyter Hub.
Jupyter Hub is a system which allows users to access compute resources, without having to maintain a system themselves, using Jupyter Notebooks. https://jupyter.org/hub
The primary task will be to deploy a production ready Jupyter Hub service accessible via Federal ID login or via IRIS’s IAM service which allows execution within virtual machines or Kubernetes within the Cloud.
Distributed Computing Infrastructure Group - APEL Accounting Dashboards
APEL is a resource usage accounting system that collects, aggregates and stores compute, storage, and cloud usage metrics from all the resource centres of the Worldwide LHC Computing Grid and European Open Science Cloud -hub infrastructures.
We’re extending APEL to provide resources accounting to the IRIS and EU SeaDataCloud projects and for this we will need accounting dashboards to display the data.
Your role is to develop new Grafana-based dashboards, interfacing them with the MySQL and ElasticSearch based repositories of accounting data, and modifying them in response to user requirements.
Computational Mathematics Group FitBenchmarking
FitBenchmarking is an open source Python package that was first developed to compare how different fit minimizers perform on various fitting problems. The original version was designed to work with the widely adopted Neutron data reduction software Mantid. FitBenchmarking has been extended to include Scipy minimizers, and an ISIS summer student is currently extending the functionality further.
The main goal of this project will be further extending FitBenchmarking by adding a few data fitting engines into FitBenchmarking. This will allow a side-by-side comparison between packages currently used in ISIS and Diamond. In turn, this knowledge will allow users to identify efficient engines for the generation of large machine learning training datasets, helping the scientists make the most effective use of the processing power available.
Dynamic Infrastructure Group - Data Analysis as a Service
The DAaaS (Data Analysis as a Service) team builds and runs the DAaaS service for the world class scientific facilities at STFC. The aim of this service is to increase the productivity of scientists by providing them with preconfigured Virtual Machines with the software and data they need installed and readily available so they can focus on analysing their data. We have IDAaaS (ISIS DAaaS) running in production for a few ISIS instrument groups; and are planning to expand to Diamond and CLF.
Your primary focus will be making the DMS and DTS (Data Movement & Transfer Services – both written in Python) production ready and improve their functionality to handle read/write operations as currently read-only is supported. This service will allow DAaaS to be hosted on different sites by enabling quick transfer of experiment and user data. For example, we are planning to branch out and run DAaaS on more clouds (first over in ISIS, then likely those that are part of the IRIS project).
DAFNI Group – Pilot Projects
The DAFNI (Data and Analytics Facility for National Infrastructure) project is creating a computing platform for modelling, simulation and visualisation of UK critical infrastructure such as transport, utilities (electricity, water, broadband), housing etc.
You will be part of the DAFNI Pilot team. The work will involve working with academic, government and commercial partners to port models onto the DAFNI platform. They will be presented with challenges when containerising existing models applications, decoupling data assets and learning how to achieve scale as part of the porting process. Working with the front end developer, they will learn how to present results and present findings within the DAFNI team setting, as well as to external pilot partners.
