D.D. Grossman and Silverman H.F. Placement of records on a secondary storage device to minimize access time. Journal of the ACM (JACM), 20(3):429--438, July 1973.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....[McKusick84] As Figure 3 6 shows, randomizing placement within a disk results in high seek distances; some other intra disk placement is needed. Seek distance can be reduced using an intra disk placement algorithm called the organ pipe cylinder optimization introduced by Grossman and Harvey [Grossman73] and recently pursued by Carson and Vongsathorn [Carson90] and Ruemmler [Ruemmler90] In this algorithm, cylinders (or tracks or blocks) are sorted according to access frequency. 37 The most heavily used block is placed in the center of the disk. The second and third most heavily used blocks are ....

David D. Grossman and Harvey F. Silverman. Placement of records on a secondary storage device to minimize access time. JACM, 20(3):429--38, July 1973.

....on the techniques used to assign files to locations on the disk. The most important result from work on the assignment problem was the organ pipe heuristic which states that; given a known, fixed distribution of disk accesses; the optimal data layout follows a profile like a set of organ pipes [6, 26]. Frequently accessed data should be place near the center of the disk just as the larger organ pipes are located in the center of the layout. Less frequently accessed data is placed towards the inner and outer edges of the disk. Unfortunately, data references are not accurately modeled as known, ....

D. D. Grossman and H. F. Silverman. Placement of records on a secondary storage device to minimize access time. Journal of the ACM, 20(3):429--438, 1973.

....disk, reducing the expected seek time. The rotational delays may also be reduced, if it is likelier that accesses will be to the same cylinder. Placing active data near the center of a disk is not a new idea. In 1973, it was shown that placement of active data in the center of the disk is optimal [58, 122], but how this optimal placement would be achieved in real systems was not discussed. For years, some system administrators have been placing a few of their hottest files in the center manually [9] iPcress automates this process, and Chapter 7 reports the performance gains. Such reorganization ....

D. D. Grossman and H. F. Silverman. Placement of records on a secondary storage device to minimize access time. Journal of the ACM, 20(3):429--438, July 1973.

....le to be sorted is stored on a sequential tape, a process of loading blocks of records into internal memory where they are sorted and using the tape to merge the sorted blocks turns out quite naturally to be more ecient than running a sorting algorithm on the entire le. Grossman and Silverman [GS73] considered the very general problem of storing records on a secondary storage device to minimize expected retrieval time, when the probability of accessing any record is known in advance. The authors model the pattern of accesses by means of a parameter that characterizes the degree to which the ....

D. D. Grossman and H. F. Silverman. Placement of records on a secondary storage device to minimize access time. Journal of the ACM, 20(3):429-438, July 1973.

....1.1 Related Work That a disk s performance can be improved by clustering frequently accessed data is well known. If data references are derived from an independent random process with a known, fixed distribution, it has been shown that the organ pipe heuristic places the data optimally [Wong 80, Grossman 73] The organ pipe heuristic calls for the most frequently accessed data to be placed in the center of the disk. The next most frequently accessed data is placed to either side of the center, and the process continues until the leastaccessed data has been placed at the edge of the disk. More ....

Grossman, David D., Harvey F. Silverman, "Placement of Records on a Secondary Storage Device to Minimize Access Time," JACM, Vol.20, No.3, July 1973.

....in subsequent studies of the performance of replication in the presence of updates. 1.1 Related Work Reducing the seek delay in disks has been the subject of many studies over the last two decades. A large group of studies have dealt with the placement of data on disk for optimizing seek times [Grossman 73, Wong 80, Carson 89] In [Carson 89] an adaptive method for reorganizing data according to the changing access patterns is studied. This study suggests that adaptive reorganization of data can outperform static placement techniques. There have also been many studies of disk head scheduling to ....

Grossman, David D., Harvey F. Silverman, "Placement of Records on a Secondary Storage Device to Minimize Access Time," JACM, Vol.20, No.3, July 1973.

....1.1 Related Work That a disk s performance can be improved by clustering frequently accessed data is well known. If data references are derived from an independent random process with a known, fixed distribution, it has been shown that the organ pipe heuristic places the data optimally [Wong 80, Grossman 73] The organ pipe heuristic calls for the most frequently accessed data to be placed in the center of the disk. The next most frequently accessed data is placed to either side of the center, and the process continues until the leastaccessed data has been placed at the edge of the disk. More ....

Grossman, David D., Harvey F. Silverman, "Placement of Records on a Secondary Storage Device to Minimize Access Time," JACM, Vol.20, No.3, July 1973.

....half of the track. In addition, the transfer cost, and thus the average transfer cost, is constant due to the constant sector read time. So, the expected random access cost that an optimal data placement should minimize, is the seek cost. The solution for the optimal data placement for CAV disks ([6, 7, 14, 15, 16]) consists of the following arrangement: The object set O 1 contains the most frequently accessed objects which fit in one track. O 1 should be placed in the middle track of the disk. O 2 is defined similarly and should be placed next to O 1 etc. This object arrangement, given their access ....

D.D. Grossman and H.F. Silverman, "Placement of Records on a Secondary Storage Device to Minimize Access Time", Journal of the Association for Computing Machinery, Vol. 20, No. 3, July 1973, pp. 429-438.

....of data in order to achieve optimality, since now we can also vary the data transfer time by varying the placement of blocks. With traditional magnetic disks, the transfer time was constant and so was the capacity of each cylinder. Most previous work on optimal placement of blocks on disks [1, 2, 6] investigates traditional magnetic disks with uniform cylinder capacity and transfer rates. Also, the optimality criterion is to minimize either the disk head travel time or the average seek time. The work by Ford [5] looks at fixed size block placement on optical disks with varying capacity and ....

....based entirely on seek. 2. Non Toeplitz: c i;j 6= f(jj Gamma ij) The cost function is not just dependent on the distance the head travels but also on the location of the block. If the transfer time was not considered in the cost then a purely seek based cost function follows the Toeplitz form [6]. 3. Monotonicity: i j k or i j k = c i;j c i;k Monotonicity does not hold in the general case, but for most practical parameters of seek and disk capacity it holds. The optimal arrangement that minimizes expected cost has the following properties: 1) Consecutivity: The blocks on a ....

David D. Grossman and Harvey F. Silverman. Placement of records on a secondary storage device to minimize access time. JACM, pages 429--438, July 1973.

....Research supported in part by an NSF Postdoctoral Fellowship. E Mail: waarts cs.berkeley.edu 1 Introduction Motivation There has been a considerable amount of work on the problem of optimizing the arrangement of data on the tracks cylinders of a disk so as to minimize the seek time (cf. [B, GS, HLP, KMW, P, YW, W]) All these papers assume a simple probability distributions for access sequences and hence aim at finding a static arrangement that is optimal for this probability distribution. One immediate problem with the approach above is determining the appropriate distribution. Moreover, even if one is ....

D.D. Grossman and H.F. Silverman, "Placement of records on a secondary storage device to minimize access time", JACM, 20 (1973), pp. 429--438.

No context found.

D.D. Grossman and Silverman H.F. Placement of records on a secondary storage device to minimize access time. Journal of the ACM (JACM), 20(3):429--438, July 1973.

No context found.

D.D. Grossman and H.F. Silverman, "Placement of records on a secondary storage device to minimize access time", JACM, 20 (1973), pp. 429--438.

No context found.

D. D. Grossman and H. F. Silverman. Placement of records on a secondary storage device to minimize access time. Journal of the ACM, 20(3):429-438, July 1973.

No context found.

David D. Grossman and Harrey F. Silverman "Placement of Records on a Secondary Storage Device to Minimize Access Time", Journal of the ACM, 20(3), 429-438 (1973).

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