H. C. Du, Disk Allocation Methods for Binary Cartesian Product Files, BIT,Vol. 26, 1986, pp. 138-147.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....notguar) teed [2] If all of the values of p j set ar selected as 1, the GDM allocation method isrD duced to the DM allocation method. It has been shown that the DM ismor e#ective than the RDA, and the perD(9#C4( of the GDM isgener)4# super#R to that of the DM. TheBinar Disk Modulo (BDM) method [3] assumes that the value of m is a power of 2, and attr #DRM valuesar binar . The equation of assigning a bucket to the disk is the same as GDM, but the value of p j set is defined as p j =2 (j mod log 2 m) The BDM s super#39D y to the DM inter4 ofstrDD optimality and in its rs# onse time ....

....of m is a power of 2, and attr #DRM valuesar binar . The equation of assigning a bucket to the disk is the same as GDM, but the value of p j set is defined as p j =2 (j mod log 2 m) The BDM s super#39D y to the DM inter4 ofstrDD optimality and in its rs# onse time toquerR3) is alsopr ved in [3]. The field exclusive allocation method, FX method, waspr9 posed in [9] A perG34#CD) analysis shows that this apprDR hgener3#C pr vides better perDRM#CRD than the GDM method. Recently, the ECC method which has the main idea ofgr3G(44 the buckets in such a way that eachgrDN forN anerR4 corGMD#C4 ....

H.C. Du,"Disk allocation methods for binary cartesian product files," BIT, vol.26, pp.138--147, 1986.

....time for an individual query and increases the overall throughput of systems. Declustering has its origins in the concept of horizontal partitioning initially developed as a distribution mechanism for distributed DBMS [19] Since then, the declustering problem has received extensive attention [1, 2, 3, 4, 5, 6, 7, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18]. In recent years, many researchers have concentrated on declustering techniques for parallel database machines. Three kinds of declustering methods have been applied in well known parallel database machines, namely round robin [20] hashed declustering [21] and range partition declustering [22] ....

....range queries. Thus, multidimensional declustering methods need additional research. However, only a little attention has been focused on this area so far. In 1982, Du and Sobolewski proposed a heuristic data allocation method, called Disk Modulo [1] which has been explored by many researchers [2, 3, 4, 5, 6]. The method considered the multidimensional access method, but was restricted to static files and partial match queries only. In 1987, Wu and Burkhard proposed a dynamic file allocation method [7] called M cycle allocation scheme, which was the first to adapt dynamic hash files to process ....

H. C. Du, Disk Allocation Methods for Binary Cartesian Product Files, BIT, Vol. 26, 1986, pp. 138-147.

....on any of the N disks. An allocation method is strictly optimal with respect to a query set Q s = fq 1 ; q 2 ; q k g if and only if the allocation method is strictly optimal for all q i , with i = 1,2, k) We note that no methods can achieve strict optimality for all query sets [7, 18, 36]. For example, no method can be strictly optimal for all range queries if the number of disks is greater than 5 [18, 36] However, some of the existing declustering methods have been proved to be strictly optimal for simpler query sets [8, 11, 36] We use a weighted similarity graph to capture ....

H. C. Du. "Disk Allocation Methods for Binary Cartesian Product Files". BIT, 26:138--147, 1986.

....items need to be accessed on any of the N disks. An allocation method is strictly optimal with respect to a query set Q s = fq 1 ; q 2 ; q k g if and only if the allocation method is strictly optimal for all q i , with i = 1,2, k) Definition 4 is similar to that used by Du et. al [7, 18, 11, 35]. We note that no methods can achieve strict optimality for all query sets [7, 18, 35] For example, no method can be strictly optimal for all range queries if the number of disks is greater than 5 [18, 35] However, some of the existing declustering methods have been proved to be strictly optimal ....

....with respect to a query set Q s = fq 1 ; q 2 ; q k g if and only if the allocation method is strictly optimal for all q i , with i = 1,2, k) Definition 4 is similar to that used by Du et. al [7, 18, 11, 35] We note that no methods can achieve strict optimality for all query sets [7, 18, 35]. For example, no method can be strictly optimal for all range queries if the number of disks is greater than 5 [18, 35] However, some of the existing declustering methods have been proved to be strictly optimal for simpler query sets [8, 11, 35] We use a weighted similarity graph to capture ....

H. C. Du. "Disk Allocation Methods for Binary Cartesian Product Files". BIT, 26:138--147, 1986.

....to the data. Simulation results for various environments comparing different replication strategies are also provided. Keywords: Data Replication, Heterogeneous Disk Systems 1 Introduction Declustering data on parallel disk systems, is a well known technique for improving I O performance [3, 5, 7, 8, 6, 9]. The idea is to distribute the data among n parallel disks so that data which is likely to be requested together by a query This work was supported by the Office of Energy Research, Office of Computational and Technology Research, Division of Mathematical, Information, and Computational ....

H. C. Du. Disk Allocation Methods for Binary Cartesian Product Files. BIT, 26, pp. 138-147, 1986.

....chips (or in general I O devices) In general, the problem of allocating a two dimensional grid on multiple I O devices can be viewed as an important instance of range queries for multidimensional files. Two types of queries have been considered: partial match queries and range queries [AU79,Du86,Sun87,KP88,FM91,AE93] The problem discussed in this paper can be viewed as a special case of range queries that are limited to two attributes on a relation with n attributes (the two attributes are the x and y axes, and the rest of the attributes correspond to the data associated with each ....

H. C. Du. Disk allocation method for binary cartesian product files. BIT, 26(2):138--147, 1986.

.... Symbol Definition M number of disks k number of attributes D i domain of i th attribute d i number of ranges of i th domain diskOf( function that maps bucket ids to disks Table 1: Symbols and Definitions Notice that there need not exist a strictly optimal allocation method for a given file [7]. Table 1 contains a list of mathematical symbols and their definitions. Given the above definitions, our goal can be described more formally as follows: Problem Definition: Given a cartesian product file with k attributes and domains D 1 ; D 2 ; D k M units (e.g, disks) Assign ....

....domain is divided into 8 ranges (d 1 = d 2 = 8) Thus, relation R consists of 64 buckets. Figure 2 shows how Disk Modulo Method would allocate these buckets to M=4 disks. Derivatives of the Disk Modulo method include the Generalized Disk Modulo allocation method and the Binary Disk Modulo method [7]; a similar approach is followed in [3] For the rest of this paper we shall use the Disk Modulo method, because it is simpler than the rest of the Modulo allocation methods and because it requires no restrictions on the number of disks or the cardinalities of the attributes. 2.2 Field wise ....

[Article contains additional citation context not shown here]

H.C. Du. Disk allocation methods for binary cartesian product files. BIT, 26:138--147, 1986.

No context found.

H. C. Du, Disk Allocation Methods for Binary Cartesian Product Files, BIT,Vol. 26, 1986, pp. 138-147.

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