Data Domain Reception Extension
Organizers: DataDomain
Thurday, December 15, 8:00 p.m.–9:00 p.m., Oregon
After a brief presentation on how Data Domain is doing for backup and
disaster recovery what the iPod did for music, there will be an even
shorter overview of career opportunities. There will also be plenty
of refreshments and a chance to meet some of the Data Domain team.
File System Benchmarking and Tools: Past, Present, and Future
Organizers: Erez Zadok, Stony Brook University; Spencer Shepler, Sun
Wednesday, December 14, 7:00 p.m.–9:00 p.m., Washington/
California
File System Benchmarking is especially difficult, for two reasons. (1)
Complex interactions between I/O devices, specialized caches, kernel
daemons, and other operating system components result in behavior that is
difficult to analyze. (2) each file system has its own features and is
optimized for certain conditions and workloads, so there is no single,
uniform way to benchmark every file system. In this BoF, we first present a
survey of file system benchmarks used in 68 recent research papers,
highlighting the pros and cons of techniques used in these papers. As you
will see, there are many flaws in current tools and techniques used to
conduct file system benchmarking.
We believe that for file system benchmarking to be better, two things are
sorely needed. First, we need a good, flexible, extensible, and
standardized file system benchmarking tool-suite; to that effect, we will
discuss a new suite under development called
FileBench. Second,
we need a centralize repository of file system traces that would be
accessible to researchers in the community (but that's the subject for
another BoF :-)
The BoF slides for the File System Benchmarking and FileBench parts (both in
PDF) are now available.
I/O Traces, Tools and Analysis
Organizers: Kristal Pollack, UC Santa Cruz;
Alistair Veitch, HP Labs
Wednesday, December 14, 9:00 p.m.–10:00 p.m., Nevada
One of the biggest challenges in testing new system designs is finding
relevant traces for performing realistic experiments. The traces
available for researchers today are out-dated and often require
specialized tools that are out-dated as well. We plan to discuss a new
"World-Wide Repository" provided by the SNIA that will be used to pool I/O
traces and tools from companies and organizations around the world. The
discussion will include standard formats and semantics for I/O traces that
will be used in this repository.
The slides (PDF) for this BoF are now available.
Sun Vendor BoF
ZFS: The Most Advanced File Systems on the Planet
Organizers: Jeff Bonwick, Distinguished Engineer, Sun Microsystems;
Bill Moore, Senior Staff Engineer, Sun Microsystems
Wednesday, December 14, 9:00 p.m.–11:00 p.m., Washington/
California
Come join the Solaris ZFS development team as they present ZFS
at FAST '05! The team recently delivered a revolutionary file
system that provides simple administration, transactional semantics,
end-to-end data integrity, immense scalability, and excellent
performance.
Don't miss the chance to be part of the various discussions on design
philosophies behind ZFS and ideas on future enhancements!
Need help? Use our Contacts page.
