Files in this item
|(no description provided)|
|Title:||Pragmatics of File Migration|
|Author(s):||Heuring, Gerald Raymond|
|Doctoral Committee Chair(s):||Lawrie, Duncan H.,|
|Department / Program:||Computer Science|
|Degree Granting Institution:||University of Illinois at Urbana-Champaign|
|Abstract:||File migration is the movement of files from secondary storage devices, such as magnetic disks, to tertiary or mass storage devices, such as automated tape libraries. It provides a service analogous to demand paging in virtual memory systems.
Previously, file migration systems have been set up using guesses and past experience to estimate the operational parameters of systems. This is error prone and does not provide a good estimate of the speed and capacity required of a tertiary storage device. The alternative was to gather large amounts file usage data over an extended period of time and use simulations on this data to provide system parameters. Using data collected from existing systems and multivariate statistics a method of sampling data from an existing system is given and these results are compared with known values.
The policies used to select which files are to be migrated from secondary to tertiary storage have been investigated by several groups and this thesis re-evaluates the stochastically optimal policy presented originally by Alan Jay Smith. Differences between the underlying simulators used at the University of California at Berkeley and the University of Illinois at Urbana-Champaign are discussed and the results of using the stochastically optimal policy on the data obtained from the University of Illinois' computer systems are presented.
Most previous efforts evaluating file migration systems have concentrated on selection of files for migration. The management of the media within the tertiary storage device is also of importance. Methods of assigning files to media, combining partially empty media, and selecting media for removal from the tertiary storage system are investigated and results presented. In addition, the cost trade-offs between using a read/write medium type versus a write-once medium type are considered along with the integration of file backup and migration.
A recommended set of steps for establishing a file migration system are also given combining the results obtained by this investigation with previous investigations into file migration.
Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 1993.
|Date Available in IDEALS:||2014-12-17|