B. McNutt. Background data movement in a log-structured disk subsystem. IBM Journal of Research and Development, 38(1):47-58, 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....defragmentation operation called cleaning. Ideally, all necessary cleaning would be completed during idle time, but this is not always possible in a busy system. The potential and actual penalties associated with cleaning have been the subject of heated debate [50] and several research e orts [52, 37, 7, 39, 59]. With freeblock scheduling, the cost of segment cleaning can be close to zero for many workloads. 5.1 Design Cleaning of a previously written segment involves identifying the subset of live blocks, reading them into memory, and writing them into the next segment. Live blocks are those that have ....

Bruce McNutt. Background data movement in a logstructured disk subsystem. IBM Journal of Research and Development, 38(1):47-58, 1994.

....means that the garbage collector is doing a good job of collecting free space, at an average of C(1 Gamma GCU) per collected segment. GCU is related to other performance measures that have been considered elsewhere. The write cost of [10] is equal to 2= 1 Gamma GCU) and the moves per write of [7] is equal to GCU= 1 Gamma GCU) Both of these are increasing functions of GCU, so that the relative ordering of different algorithms by GCU is the same ordering that would be obtained using either of the other measures. Note also that write cost and moves per write both approach infinity as GCU ....

B. McNutt, Background data movement in a log-structured disk subsystem, IBM Journal of Research and Development 38, 1 (Jan. 1994), pp. 47--58.

....is doing a good job of collecting free space, at an average of C(1 Gamma GCU) per collected segment. GCU is related to other performance measures that have been considered elsewhere. The write cost of (Rosenblum and Ousterhout, 1992) is equal to 2= 1 Gamma GCU) and the moves per write of (McNutt, 1994) is equal to GCU= 1 Gamma GCU) Both of these are increasing functions of GCU, so that the relative ordering of different algorithms by GCU is the same ordering that would be obtained using either of the other measures. Previous studies of garbage collection algorithms show (not surprisingly) ....

McNutt, B. (1994). Background data movement in a log-structured disk subsystem.

.... indicated that LFS provided better write performance than conventional file systems and comparable read performance [2] Subsequent work re evaluated LFS and concluded that, with garbage collection, the performance of LFS on secondary storage is only comparable to conventional file systems [5, 6, 7] With transaction processing workloads, in particular, the throughput of LFS is lower than expected [8] HighLight [9] a tertiary storage extension of BSD LFS, is a first attempt in combining LFS with tertiary storage. HighLight migrates segments of user data and meta data from secondary storage ....

B. McNutt. Background data movement in a logstructured disk subsystem. IBM J. Res. Develop., 38(1):47--58, January 1994.

.... Fortunately, studies on I O access patterns, disk shuffling, and file system restructuring have shown that these conditions are often met in practice [Akyurek and Salem 1993; Deshpandee and Bunt 1988; Floyd and Schlatter Ellis 1989; Geist et al. 1994; Majumdar 1984; McDonald and Bunt 1989; McNutt 1994; Ruemmler and Wilkes 1991; Ruemmler and Wilkes 1993; Smith 1981] Such a storage hierarchy could be implemented in a number of different ways: Manually, by the system administrator. This is how large mainframes have been run for decades. Gelb 1989] discusses a slightly refined version of this ....

.... HP AutoRAID owes an intellectual debt to the body of work on log structured file systems (LFS) Carson and Setia 1992; Ousterhout and Douglis 1989; Rosenblum and Ousterhout 1992; Seltzer et al. 1993; Seltzer et al. 1995] and cleaning (garbage collection) policies for them [Blackwell et al. 1995; McNutt 1994; Mogi and Kitsuregawa 1994] There is a large literature on hierarchical storage systems and the many commercial products in this domain (for example [Chen 1973; Cohen et al. 1989; DEC 1993; Deshpandee and Bunt 1988; Epoch Systems Inc. 1988; Gelb 1989; Henderson and Poston 1989; Katz et al. 1991; ....

MCNUTT, B. 1994. Background data movement in a log-structured disk subsystem. IBM J. Res. and Development 38, 1. IBM Corp., Armonk, New York, 47--58.

.... the active subset must change relatively slowly over time (to allow the array to do useful work, rather than just move data between the two levels) Fortunately, studies on I O access patterns, disk shuffling and file system restructuring have shown that these conditions are often met in practice [Akyurek93, Deshpande88, Floyd89, Geist94, Majumdar84, McDonald89, McNutt94, Ruemmler91, Ruemmler93, Smith81]. Such a storage hierarchy could be implemented in a number of different ways: Manually, by the system administrator. This is how large mainframes have been run for decades. Gelb89] discusses a slightly refined version of this basic idea. The advantage of this approach is that human ....

....approach adopted in HP AutoRAID cost effective. The log structured writing scheme used in HP AutoRAID owes an intellectual debt to the body of work on log structured file systems (LFS) Carson92, Ousterhout89, Rosenblum92, Seltzer93, Seltzer95] and cleaning (garbage collection) policies for them [McNutt94, Blackwell95]. There is a large literature on hierarchical storage systems and the many commercial products in this domain (for example [Chen73, Cohen89, DEC93, Deshpande88, Epoch88, Gelb89, Henderson89, Katz91, Miller91, Misra81, Sienknecht94, Smith81] together with much of the proceedings of the IEEE ....

Bruce McNutt. Background data movement in a logstructured disk subsystem. IBM Journal of Research and Development, 38(1):47--58, 1994.

No context found.

B. McNutt. Background data movement in a log-structured disk subsystem. IBM Journal of Research and Development, 38(1):47-58, 1994.

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