Initial Phase of SAGE Project Successfully Completed

Capacity and performance requirements for future storage architectures can hardly be met by one single storage device technology. While hard drives continue to provide more capacity, they remain slow. Storage devices based on non-volatile memory are much faster, but also much more expensive. Hierarchical storage architectures with tiers optimized for capacity and others optimized for performance allow us to solve this dilemma. While the hardware technologies are available, integrated architectures including a suitable software layer are still lacking. The EC-funded project SAGE (Percipient StorAGe for Exascale data centric computing) addresses this gap.

At the beginning of June the project was subject to an interim review. The reviewers highlighted the good progress that had been made during the first 9 months of the three-year project. In this period, the focus was on establishing a co-design process and deriving requirements for the final architecture for different workloads. SAGE’s application portfolio comprises a broad range of applications from a variety of research fields, which share the common aspect that they are coping or will have to cope with data volumes at extreme scales, as their problem domain scales upward.

The identified requirements are now used as input for the different R&D tasks within SAGE. In parallel, the application experts in the project will start to develop strategies for refactoring applications in order to efficiently exploit the features of SAGE. While there will be support for legacy codes, applications will have to be adapted in order to optimally exploit hierarchical storage architectures. Further information at http://sagestorage.eu/.
(Contact: Prof. Dirk Pleiter)

JSC News No. 242, 16 June 2016