Cloud Computing Service
A comprehensive cloud computing service, to provide a platform to self-provision infrastructure to support both researchers and development teams.
Information Services is providing an interface to the the University’s research computing infrastructure, to provide support to researchers who require a platform to self-build their own server infrastructure and utilise the storage from Datastore.
Information Services is also planning a cloud service to provide an agile development platform and a burst capacity service to support developers in IS and schools and to provide a stable service with a higher guarantee of reliability than that of the research service.
These services are implemented using industry standard OpenStack software.
Researcher Cloud Service
This service is now available for research use.
These use cases will guide researchers to using the Research Cloud Service:
a researcher needs access to an API to create virtual machines to run compute jobs accessing research data on Datastore or the Secure Data Service using operating systems not provided on the standard Eddie compute platform;
access to an API to create virtual machines with the ability to use pre-defined research-related applications and operating system environments which have been pre-packaged and provided as a “cloud image”;
access to scheduling tools which are fundamentally different from GridEngine – in particular, Spark and Hadoop;
the ability to build and deploy custom workflow and other research management tools which may include data management, data processing, simulation, analysis and publication as a single packaged pipeline;
provide bespoke interfaces to research data and tools for sharing and collaborating across and outside the University of Edinburgh, both with named collaborators and the general public.
Accessing the service
Developer Cloud Service
This service will be made available later in the summer of 2017.
These use cases will guide people to using the Developer Cloud service:
Agile development use cases
- a developer needs to quickly spin up a new instance of a virtual machine to develop a new service, test and then destroy it again and repeat, without the overhead of having to request the creation of a virtual machine through the virtual hosting service. This would support an agile development process and complement a DevOps approach to service development;
an application developed as above needs to be migrated to the virtual hosting service when it enters full production service. Running on a common platform with the virtual hosting service will enable an easier migration path for the virtual machine image to be moved into full production.
- an application that routinely runs on a small virtual machine or small number of virtual machines on the virtual hosting service but occasionally needs to “burst out” and replicate itself to provide greater processing capacity during busy periods;
an application running on a non-virtualised service but that can be pre-packaged as a “cloud image” and can be spun up through an API to provide extra processing capacity during busy periods similar to the above. An example, of such a service might be Edina’s Digimap for Schools service;
an application developed through the agile development process above needs to enter production as a scalable service from the outset, running up additional instances as required.