We present a policy-based architecture STEPS for lifecycle management (LCM) in a mass scale distributed file system. The STEPS architecture is designed in the context of IBM’s ...
Akshat Verma, David Pease, Upendra Sharma, Marc Ka...
Many parallel scientific applications need high-performance I/O. Unfortunately, end-to-end parallel-I/O performance has not been able to keep up with substantial improvements in p...
We describe an approach to verifying bit-level pipelined machine models using a combination of deductive reasoning and decision procedures. While theorem proving systems such as AC...
Reliable storage systems depend in part on “write-before” relationships where some changes to stable storage are delayed until other changes commit. A journaled file system, ...
Christopher Frost, Mike Mammarella, Eddie Kohler, ...
Unchecked errors are especially pernicious in operating system file management code. Transient or permanent hardware failures are inevitable, and error-management bugs at the fi...