 |
7th Parallel Data Storage Workshopheld in conjunction with
|
KEYNOTE SPEAKER: Eric Barton, Intel
Fast Forward Storage and I/O
Abstract: Conflicting pressures drive the requirements for I/O and Storage at Exascale. On the one hand, an explosion is anticipated, not only in the size of scientific data models but also in their complexity and in volume of their attendant metadata. These models will require new object-oriented I/O APIs to make application development tractable and create new workflows that integrate analysis and visualisation and move the compute to the data or the data to the compute as appropriate. On the other hand, economic realities driving the architecture and reliability of the underlying hardware will push the limits on horizontal scale, introduce unavoidable jitter and make failure the norm. The I/O system will have to handle these as transparently as possible while providing efficient, sustained and predictable performance. This talk will describe the research underway in the DOE Fast Forward project to prototype a complete Exascale I/O stack including at the top level, an object-oriented I/O API based on HDF5, in the middle, a Burst Buffer and data layout optimizer based on PLFS and at the bottom, DAOS – transactional object storage based on Lustre.
Bio: Eric Barton is the lead architect of the High Performance Data Division at Intel. He received his BSc in Computer Science at Edinburgh University in 1979 and then studied at the California Institute of Technology, focusing on CAD and parallel processing. He curtailed these studies to join Inmos where he helped to implement the silicon design system which was used to create the Transputer. In 1985 Eric co-founded Meiko Ltd to build massively parallel systems based first on the Transputer, then on Sparc using Transputer networks as the communications fabric. The final generation of Meiko computers, based on Sparc plus Elan/Elite had a scalable RDMA fabric around which Eric's team built a complete parallel programming environment including a parallel job scheduler, communications libraries, I/O staging libraries and a parallel file system. In 1996 Eric left Meiko to work as an independent consultant and in 2002 was engaged by Cluster File Systems to work on Lustre. He continued to work on Lustre, becoming Lead Engineer of the Lustre Group after Sun Microsystems acquired CFS and in 2010, Eric co-founded Whamcloud, Inc. to assure the continued development and support of Lustre as an Open Source GPL filesystem.