Working with OCF and IBM, the University of Birmingham Research Computing Team has selected IBM’s Spectrum Scale (formerly GPFS) Data Management Edition to underpin it’s research data storage systems.
Dr John Owen, Head of Research Support at the University said, “Looking after research data is key to our researchers. The value of the research we undertake is locked away in the data and we need to ensure that our researchers have safe, secure and reliable storage to keep it”.
Research computing at the University provides a wide range of advanced computing facilities which are free for researchers to use, from the BlueBEAR HPC system, BEARCloud research cloud platform, research data services through to high performance networking to connect research equipment to the storage. At the heart of the platforms is ensuring data is looked after.
Simon Thompson, Research Computing Infrastructure Architect says, “We’ve been a long term user of Spectrum Scale for our storage systems, and it’s a very stable platform for us. We use it for both data storage and as part of our private cloud deployment – for storing VM images, it means we have a single data management plane and easily allows us to place data into different classes of storage depending on the need for access – for example with data we need to archive, we can move it to tape via the Spectrum Protect integration where users can seamlessly restore it. One of the key features of Spectrum Scale for us is that we can use different classes and vendors of storage underneath it so that we can performance optimise placement depending on the required workload.”
He continues “As we bring in more research data from local storage, we have increasing requirements for encryption, and Data Management Edition provides us with a certified platform that allows us to support encrypted data at rest. We’ll be using placement policies to define which classes of data need to be encrypted, but for our users it will be transparent.”
Data Management edition brings a new license model to Spectrum Scale and is managed by capacity of the system. As the University has grown the number of systems and amount of storage, this has made the current licenses difficult to manage. Simon continues, “Previously we had to account for client and server licenses and with a mix of different vendor and OEM licenses, this was proving difficult, in addition to this we were often designing for optimal license use, rather than the best architectural solution. With Data Management Edition, we no longer need to worry about server and client licenses, which is particularly of interest for some of the environments we are looking to develop in our private cloud infrastructure.”
Looking to the future, the University is keen to explore other features of Data Management Edition, for example “transparent cloud tiering”, which allows automatic migration of data into object storage is particularly of interest.
Working with OCF and the IBM sales team, the University was able to migrate their existing IBM standard licenses over to the new license model.