Listen in MP3 format

A System's Hackers Crash Course: Techniques that Find Lots of Bugs in Real (Storage) System Code
Dawson Engler, Professor, Stanford University

View the presentation slides

Listen in MP3 format

This talk describes several effective bug-finding tools we have developed, which exploit not-widely-understood techniques—implementation-level model checking and symbolic execution—focusing on the key intuitions and ideas behind them.

These tools have found errors in every system we checked, including: three version control systems, BerkelyDB, an NFS implementation, ten file systems, a RAID system, and the popular VMware GSX virtual machine. These errors have been serious, such as where an inopportune crash will cause various Linux file systems (e.g., ext3) to to trash their entire root directory "/" or bad disk images that, when mounted, will crash the system.

The talk will close with some of the weird surprises that happen when academics try to commercialize bug-finding research.

Dawson Engler is an Associate Professor in CS and EE at Stanford. He received his PhD from MIT for his work on the exokernel operating system and his undergraduate degree from University of Arizona, the latter in large part funded by being a bouncer. His research focuses on developing techniques to find interesting bugs in real code, including static analysis, implementation level model checking, and symbolic execution. His research group has won numerous "Best Paper" awards and its static tools have found thousands of errors in open source systems (e.g., Linux and BSD) and have formed the basis of a company, Coverity, which has over 200 customers and 70 employees. He won the 2006 Weiser award.

Awarded Best Paper!
Disk Failures in the Real World: What Does an MTTF of 1,000,000 Hours Mean to You?
Bianca Schroeder and Garth A. Gibson, Carnegie Mellon University

Failure Trends in a Large Disk Drive Population
Eduardo Pinheiro, Wolf-Dietrich Weber, and Luiz André Barroso, Google Inc.

A Five-Year Study of File-System Metadata
Nitin Agrawal, University of Wisconsin, Madison; William J. Bolosky, John R. Douceur, and Jacob R. Lorch, Microsoft Research

Proportional-Share Scheduling for Distributed Storage Systems
Yin Wang, University of Michigan; Arif Merchant, HP Laboratories

Argon: Performance Insulation for Shared Storage Servers
Matthew Wachs, Michael Abd-El-Malek, Eno Thereska, and Gregory R. Ganger, Carnegie Mellon University

Strong Accountability for Network Storage
Aydan R. Yumerefendi and Jeffrey S. Chase, Duke University

The FAST technical sessions will include slots for Work-in-Progress reports, preliminary results, and "outrageous" opinion statements. The WiPs schedule and abstracts are now available.

Trends in Managing Data at the Petabyte Scale
Steve Kleiman, CTO, Network Appliance

View the presentation slides

Listen in MP3 format

The explosive growth in stored data has made petabyte-scale storage infrastructures increasingly common. The scale, growth rate, and increases in regulations related to data storage have imposed a number of non-obvious burdens on data ownership. These trends are driving the need to reorganize the traditional application-centric storage architectures toward a more unified storage infrastructure with new data management paradigms. This reorganization will likely drive a vibrant storage market over the next ten years.

Steve Kleiman joined Network Appliance in April 1996. He is currently senior vice president and chief technology officer and is responsible for setting future technology and product directions for the company. Kleiman has designed and developed UNIX and workstation architecture for 22 years.

He began his career in UNIX development at Bell Telephone Laboratories in 1977, where he helped develop the first x86-based UNIX product. Kleiman then moved to Sun Microsystems, where he worked from 1984 to 1996 as a Distinguished Engineer and chief architect of clustered UNIX systems. As chief technologist for Sun's Interactive Services Group, he designed the company's first video server product line. Kleiman was also lead architect for multithreading and multiprocessing in Solaris and is a member of the POSIX Pthreads committee. He developed the Vnodes file system interface and was a member of the original NFS development team at Sun. Kleiman was the project leader of the original port of SunOS to SPARC.

He received a master's degree in electrical engineering from Stanford University in 1978 and a bachelor's degree in electrical engineering and computer science from the Massachusetts Institute of Technology in 1977.

Architectures for Controller Based CDP
Guy Laden, Paula Ta-Shma, Eitan Yaffe, Michael Factor, and Shachar Fienblit, IBM Haifa Research Laboratory

Jumbo Store: Providing Efficient Incremental Upload and Versioning for a Utility Rendering Service
Kave Eshghi, Mark Lillibridge, Lawrence Wilcock, Guillaume Belrose, and Rycharde Hawkes, HP Laboratories

Data ONTAP GX: A Scalable Storage Cluster
Michael Eisler, Peter Corbett, Michael Kazar, and Daniel S. Nydick, Network Appliance; J. Christopher Wagner, IronPort Systems, Inc.

//TRACE: Parallel Trace Replay with Approximate Causal Events
Michael P. Mesnier, Intel Research with Carnegie Mellon University; Matthew Wachs, Raja R. Sambasivan, Julio Lopez, James Hendricks, Gregory R. Ganger, and David O'Hallaron, Carnegie Mellon University

Karma: Know-It-All Replacement for a Multilevel Cache
Gala Yadgar, Technion; Michael Factor, IBM Haifa Research Laboratories; Assaf Schuster, Technion

AMP: Adaptive Multi-stream Prefetching in a Shared Cache
Binny S. Gill and Luis Angel D. Bathen, IBM Almaden Research Center

Nache: Design and Implementation of a Caching Proxy for NFSv4
Ajay Gulati, Rice University; Manoj Naik and Renu Tewari, IBM Almaden Research Center

Awarded Best Paper!
TFS: A Transparent File System for Contributory Storage
James Cipar, Mark D. Corner, and Emery D. Berger, University of Massachusetts, Amherst

Cobalt: Separating Content Distribution from Authorization in Distributed File Systems
Kaushik Veeraraghavan, Andrew Myrick, and Jason Flinn, University of Michigan

PARAID: A Gear-Shifting Power-Aware RAID
Charles Weddle, Mathew Oldham, Jin Qian, and An-I Andy Wang, Florida State University; Peter Reiher, University of California, Los Angeles; Geoff Kuenning, Harvey Mudd College

REO: A Generic RAID Engine and Optimizer
Deepak Kenchammana-Hosekote, IBM Almaden Research Center; Dingshan He, Microsoft; James Lee Hafner, IBM Almaden Research Center

PRO: A Popularity-based Multi-threaded Reconstruction Optimization for RAID-Structured Storage Systems
Lei Tian and Dan Feng, Huazhong University of Science and Technology; Hong Jiang, University of Nebraska—Lincoln; Ke Zhou, Lingfang Zeng, Jianxi Chen, and Zhikun Wang, Huazhong University of Science and Technology and Wuhan National Laboratory for Optoelectronics; Zhenlei Song, Huazhong University of Science and Technology