| Masataka Ohta and Hiroshi Tezuka. A fast /tmp file system by delay mount option. 1990 Summer USENIX Technical Conference (Anaheim, California, June 1990), pages 145--50. USENIX, June 1990. |
....disk [Livny87] Given our assumption that future file servers will have multiple devices available, log structured file implementations will most likely be striped. 13 2.1. 4 Delayed write file implementations The delayed write file implementation trades off fault tolerance for performance [Ohta90]. Unix uses a write back cache to speed file accesses. To minimize data loss in the case of a failure, Unix makes two exceptions to the write back discipline. First, it periodically flushes all dirty buffers, usually every half minute. Second, it writes through critical file system structures ....
Masataka Ohta and Hiroshi Tezuka. A fast /tmp file system by delay mount option. 1990 Summer USENIX Technical Conference (Anaheim, California, June 1990), pages 145--50. USENIX, June 1990.
....is typically only held in volatile memory for short periods of time before being written to disk, this exposure is tolerated in exchange for the increased performance that results. This idea has been extended still further by special file systems that deliberately store data in volatile memory [McKusick90, Ohta90]. Most systems that use non volatile memory (NVRAM) assume that a single copy of the stored data is sufficient: providing true single failure tolerant NVRAM systems is difficult [Menon93] and so is rarely done. Examples of common systems that make this tradeoff include the popular PrestoServe ....
Masataka Ohta and Hiroshi Tezuka. A fast /tmp file system by delay mount option. 1990 Summer USENIX Technical Conference (Anaheim, California, June 1990), pages 145--50. USENIX, June 1990.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC