11th HLRS/hww Workshop on Scalable Global Parallel File Systems "Total Cost of Dataship"

The High Performance Computing Center Stuttgart, HLRS <http://www.hlrs.de/> and the Höchstleistungsrechner für Wissenschaft und Wirtschaft mbH, hww, are organizing the three-day... 

11th HLRS/hww Workshop on Scalable Global Parallel File Systems "Total Cost of Dataship" at HLRS which is one of the three federal high performance computing centers in Germany and a member of the Gauss Center for Supercomputing (GCS). This workshop will be joined by the High Performance Networking Forum (HNF) Europe Spring Meeting 2012. The workshop is to take place in Stuttgart, Germany, on Monday - Wednesday, May 7 - 9, 2012. This year's workshop will try and present the major issues and developments associated with the convergence of fabrics, protocols, systems and applications in the computing center, enabling a rich portfolio of distributed storage solutions. Focus this year is on "Total Cost of Dataship" which means handling of data in the most efficient and economic way but considering the boundary conditions of security, safety, performance, access methods, availability and other negotiated and confirmed service levels. HLRS ist looking forward to meaningful presentations coming from the following fields:

  • Scalable global parallel file systems including applications w/o OSD capability (in alphabetical order, e.g. ActiveScale FS, GFS, GPFS, Isilon Clustered Storage, Lustre, Parallel NFS, ...)
  • Linux data management middleware (e.g. Ibrix, iRods, ...)
  • Hierarchical storage systems, either standalone or by extensions to the SGPFS's above (e.g. HPSS Linux VFS, HPSS/GPFS, HSM extensions to Lustre, OSD-based storage hierarchies, Open Source Archives…)
  • New modular approaches to storage software, allowing users to supply plug-ins to an engine to achieve features analogous to e.g. SQL engine/business logic separation
  • Development of the ANSI T10 OSD Standard and its Extensions
  • Standardization efforts and Technologies for Exa-scale archives
  • Recent Data Grid and Cloud Storage Developments
  • Work on the forthcoming Data Center Ethernet (DCE)
  • Unified Wire Protocols and Subsystems
  • Fibre Channel over Ethernet, InfiniBand over Ethernet, or vice versa
  • Advanced Storage Subsystems: FC/SAS/SATA/InfiniBand RAID systems, Parallel Tapes, Smart Storage Cell employing distributed redundancy schemes, fast parity computation and restore
  • Object Storage Targets deploying migration techniques and self organization, micromechanical storage systems, high-capacity optical devices, solid-state disks …
  • Emulation of (Parallel) Tape Libraries by nearline hard disk (RAID) systems, activated on demand
  • Enhanced Monitoring and Storage Accounting
  • Interconnects: (system area: DCE, Ethernet Backplane WG, InfiniBand, (storage area: FibreChannel, Data Center Ethernet IG, Virtual SANs)

High-Performance Networking Forum (HNF) Europe sponsored sessions:

  • Development of the next generation 100-Gbit/s Ethernet standard
  • 100-Gbit/s Ethernet prototype transmission and switching systems
  • 100-Gbit/s serial host interfaces
  • 10/40-Gbit/s Ethernet and "Unified Wire" host adapters at wire rates
  • Fibre Channel over Ethernet field tests
  • InfiniBand FDR/EDR switches and host adapters
  • FDR/EDR long-range optical interfaces for InfiniBand
  • Optical Networks: (XFP multi-source agreement, 100-Gbit/s direct laser transmission, 100-Gbit/s DWDM, E/O add/drop systems, XFPs with electronic dispersion compensation, pure optical switching).