S. Navathe, S. Ceri, G. Wiederhold, and J. Dou, "Vertical Partitioning Algorithms for Database Design," ACM Transactions on Database Systems, Vol. 9 No. 4, pp. 680-710, 1984.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....traditional fashion. 2.2 The Decomposition Storage Model Vertical partitioning is the process of striping a relation into sub relations, each containing the values of a subset of the initial relation s attributes. Vertical partitioning was initially proposed in order to reduce I O related costs [26]. The fully decomposed form of vertical partitioning (one attribute per stripe) is called the decomposition storage model (DSM) 12] DSM partitions an n attribute relation vertically into n sub relations, as shown in Figure 2. Each sub relation contains two attributes, a logical record id ....

S. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical partitioning algorithms for database design. ACM Transactions on Database Systems, 9(4), pp/680-710, December 1984.

....horizontal fragmentation is reviewed followed by work on fragmentation in DOBS. Horizontal Fragmentation (relational) Several researchers have worked on fragmentation and allocation in the relational data model including Ceri, Negri and Pelagatti [2] Ozsu and Valduriez [17] Navathe et al. [13] , Navathe and Ra [15] and Shin and Irani [3] Ceri, Negri and Pelagatti [2] show that the main optimization parameter needed for horizontal fragmentation is the number of accesses performed by the application programs to different portions of data (file of records) They define applications in ....

....of minterms which form the horizontal fragments. Navathe, Karlapalem and Ra [14] define a scheme for simultaneously applying the horizontal and vertical fragmentation algorithms on a relation to produce a grid. A technique similar to the vertical fragmentation schemes discussed in Navathe et al. [13, 15] is used to produce horizontal fragments. Ozsu and Valduriez [17] define the database information needed for horizontal fragmentation of the universal relation and show how the database relations are reconstructible using joins. Ceri et al. 4] model this relationship explicitly using directed ....

S.B. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical partitioning algorithms for database design. ACM Transactions on Database Systems, 9(4), 1984.

....schemes in OODBs, and in [13] we presented a method induced approach for class partitioning and techniques for supporting fragmentation transparency. This background work facilitated in depth understanding of class partitioning in OODBs. In [1, 5] the authors adopted an affinity based approach [15] to vertical partitioning of object oriented databases. However, they did not consider any representation scheme for vertical class fragments. Also, they ignored the physical costs corresponding to the savings on the amount of irrelevant data accessed, and the overhead due to vertical ....

S. B. Navathe, S. Ceri, G. Wiederhold, and J. Dou, "Vertical partitioning algorithms for database design", in ACM Transactions on Database Systems, 9(4):680--710, 1984.

....HP because it is well adapted to data warehouses. Also, the TPC D benchmark [60] in its implementation allows the utilization of HP, but it discards the utilization of the vertical partitioning. 3. 3 Horizontal Partitioning Algorithms in Databases Several algorithms were proposed in performing HP [17, 54, 56, 55, 59, 15, 27, 26]. These algorithms can be classified into two main categories: 1) query driven algorithms and (2) datadriven algorithms. 3.3.1 Query Driven Algorithms These algorithms are performed based on a set of most frequently asked queries and their access frequencies for a specific application [55] ....

S.B. Navathe, S. Ceri, G. Wiederhold, and Dou J. Vertical partitioning algorithms for database design. ACM Transaction on Database Systems, 9(4):681--710, December 1984.

....Finally, early work on hybrid fragmentation in the relational databases is reviewed. 2.1. 1 Horizontal Fragmentation Several researchers have worked on fragmentation and allocation in the relational data model including Ceri, Negri and Pelagatti [5] Ozsu and Valduriez [30] Navathe et al. [26], Navathe and Ra [28] and Shin and Irani [6] Ceri, Negri and Pelagatti [5] show that the main optimization parameter needed for horizontal fragmentation is the number of accesses performed by the application programs to different portions of data (file of records) They characterize the ....

....of simple predicates which eliminate meaningless minterms. Navathe, Karlapalem and Ra [27] define a scheme for simultaneously applying the horizontal and vertical fragmentation algorithms on a relation to produce a grid. A technique similar to the vertical fragmentation schemes discussed in [26, 28] is used to produce horizontal fragments. They consider transactions with high access frequencies using simple or derived predicates to produce horizontal fragments. A predicate usage matrix is obtained from transaction information and leads to a predicate affinity matrix that clusters the ....

[Article contains additional citation context not shown here]

S.B. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical partitioning algorithms for database design. ACM Transactions on Database Systems, 9(4), 1984.

....from scratch, that is users are not yet assigned to subnets (see Applicability for further comments) The problem is similar to the distribution of attributes at vertical partitioning in distributed database systems. For this reason the following algorithm is adopted from this research area [12, 13]. First of all we sketch the algorithm then we explain its meaning and the meaning of the symbols used. 1. create the user step matrix use 0 ki 2. standardize use 0 ki to use ki 3. compute the user affinity matrix aff ij using use ki 4. use the known algorithms to find clusters in aff ij ....

....3 0 2 5 2 5 1 5 Now the user affinity matrix can be created. It contains the degree of the connection between users. If a cluster of users has high values in this matrix, they have common steps and should be in the same subnet. Such clusters of users can be found with the algorithms described in [12] and (with a better complexity) in [13] aff ij = P k use ki Delta use kj Delta E k Frequency of step k: k 1 2 3 E k 100 60 10 aff 1 2 3 4 1 44.4 0 0 22.2 2 0 16.6 16.6 0.8 3 0 16.6 16.6 0.8 4 22.2 0.8 0.8 11.5 In this example one cluster would consist of the users 1 and 4 and ....

S. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical Partitioning Algorithms for Database Design. ACM Transactions on Database Systems, 9(4):680--710, 1984.

....becomes more of an art than science. As argued in [21] physical database design can be automated in a large number of practical cases. However, most of the existing work focused on algorithmic design for sub problems of physical database design, such as index selection and vertical partitioning [1, 2, 3, 4, 5, 7, 14, 18, 19, 24]. Only a few work address how the various techniques should be integrated, and how the physical database design process may be automated [8, 11, 21] In this paper, we describe the design, implementation and evaluation of DATPLEX, a Design Aid Tool for Physical database design that is Layered and ....

....and storage structures and index selection. In the subsequent subsections, we present each of the layers, and the design evaluator. Due to space limitation, we can only be brief in our discussion. Generate Vertical Partitions. Several techniques for vertical partitioning have been proposed [7, 18, 19]. In [19] a graphical technique was proposed. The basic idea of the technique is to partition a relation into fragments such that all pairs of attributes within a fragment have high affinity but all attributes between fragments have low affinity . In DATPLEX, we used the same approach but the ....

[Article contains additional citation context not shown here]

S. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical partitioning algorithms for database design. ACM TODS, 9(4), December 1984.

....methods involved) and produce instance variable usage matrix Calculate instance variable affinity matrix Perform [NR89] graph based algorithm End 3. 4 Mixed Vertical Partitioning Algorithm As the exhaustive enumeration strategy used in pure cost based approach requires a high computational cost [NCWD84], it is impractical when the total number of instance variables in the schema is large. Moreover, affinity based approach is not as comprehensive as cost based approach in modeling some important database characteristics, like the size of the instance variable. Further, the affinity based approach ....

S. B. Navathe, S. Ceri, G. Wiederhold and J. Dou, "Vertical Partitioning Algorithms for Database Design", ACM Trans. on Database Systems, Vol. 9, No. 4, Dec 1984.

....This section reviews previous work on vertical fragmentation in these systems and finally covers work on fragmentation in a DOBS. Vertical Fragmentation (relational) Existing work on vertical fragmentation in the relational data model includes Hoffer and Severance [5] Navathe et al. [12], Cornell and Yu [2] Navathe and Ra [13] Ozsu and Valduriez [15] and Chakravarthy et al. 1] Hoffer and Severance [5] define an algorithm that clusters attributes of a database entity based on their affinity. Attributes accessed together by applications have high affinity so the Bond Energy ....

....and Severance [5] define an algorithm that clusters attributes of a database entity based on their affinity. Attributes accessed together by applications have high affinity so the Bond Energy Algorithm developed by McCormick et al. 11] is used to form these attribute clusters. Navathe et al. [12] extends Hoffer s work by defining algorithms for grouping attributes into overlapping and nonoverlapping fragments. The approach is to minimize the number of fragments visited by a transaction and to refine fragments using cost factors that reflect the physical environment where the fragments are ....

[Article contains additional citation context not shown here]

S.B. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical Partitioning Algorithms for Database Design. ACM Transactions on Database Systems, 9(4), 1984.

....on demand connectivity; therefore, they are not appropriate for our environment. Similar concepts have been tried in file systems for intermittent connectivity such a CODA[96] To increase scalability, we will perform data partitioning by vertical and horizontal fragmentation of relations as in [35, 119]. Some work has been done studying the prospect of self maintainable views[72] or providing additional information to facilitate view maintenance; however, since a client s views are assumed to have data in common with other views, these techniques do not apply. There are some other ....

Shamkant B. Navathe, Stefano Ceri, Gio Wiederhold, and Jinglie Dou. Vertical partitioning algorithms for database design. ACM Transactions on Database Systems, 9(4):680--710, December 1984.

....horizontally and vertically exist. Previous work on relational vertical fragmentation is reviewed and then previous work on fragmentation in DOBS. Vertical Fragmentation (relational) Work on vertical fragmentation in the relational data model includes Hoffer and Severance [6] Navathe et al. [8], Cornell and Yu [2] Ozsu and Valduriez [9] and Chakravarthy et al. 1] Hoffer and Severance [6] define an algorithm that clusters attributes of a database entity based on their affinity. Attributes accessed together by applications have high affinity so the Bond Energy Algorithm [9] is used to ....

....and Chakravarthy et al. 1] Hoffer and Severance [6] define an algorithm that clusters attributes of a database entity based on their affinity. Attributes accessed together by applications have high affinity so the Bond Energy Algorithm [9] is used to form these attribute clusters. Navathe et al. [8] extends Hoffer s work by defining algorithms for grouping attributes into overlapping and nonoverlapping fragments. Cornell and Yu [2] optimized this work by developing an algorithm that obtains an optimal binary partitioning for relational databases. Ozsu and Valduriez [9] discuss this earlier ....

[Article contains additional citation context not shown here]

S.B. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical Partitioning Algorithms for Database Design. ACM Transactions on Database Systems, 9(4), 1984.

....and Severance [4] define an algorithm that clusters attributes of a database entity based on their affinity. Attributes accessed together by applications have high affinity so the Bond Energy Algorithm developed by McCormick et al. 6] is used to form these attribute clusters. Navathe et al. [8] extends Hoffer s work by defining algorithms for grouping attributes into overlapping and Vertical Class Fragmentation in a DOBS 4 nonoverlapping fragments. Cornell and Yu [1] optimized this work by developing an algorithm that obtains an optimal binary partitioning for relational databases. Ozsu ....

....et al. 5] define issues involved in distribution design for an object oriented database system. They identify two types of methods simple and complex methods. They argue that a model consisting of simple methods can be vertically partitioned using techniques described by Navathe et al. [8], while that of complex methods, requires a method based view (MBV) The MBV identifies the set of objects accessed by a method and the set of attributes or instance variables accessed by the method. The sets are further grouped into sets of objects and instance variables based on the classes to ....

[Article contains additional citation context not shown here]

S.B. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical partitioning algorithms for database design. ACM Transactions on Database Systems, 9(4), 1984.

....relational databases. Finally, previous work on fragmentation in DOBS is reviewed. Horizontal Fragmentation (relational) Several researchers have worked on fragmentation and allocation in the relational data model including Ceri, Negri and Pelagatti [2] Ozsu and Valduriez [18] Navathe et al. [14], Navathe and Ra [16] and Shin and Irani [3] Ceri, Negri and Pelagatti [2] show that the main optimization parameter needed for horizontal fragmentation is the number of accesses performed by the application programs to different portions of data (file of records) Navathe, Karlapalem and Ra ....

....to different portions of data (file of records) Navathe, Karlapalem and Ra [15] define a scheme for simultaneously applying the horizontal and vertical fragmentation algorithms on a relation to produce a grid. A technique similar to the vertical fragmentation schemes discussed in Navathe et al. [14, 16] is used to produce horizontal fragments. Ozsu and Valduriez [18] define the database information needed for horizontal fragmentation of the universal relation and show how the database relations are reconstructible using joins. Ceri et al. 4] model this relationship explicitly using directed ....

[Article contains additional citation context not shown here]

S.B. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical partitioning algorithms for database design. ACM Transactions on Database Systems, 9(4), 1984.

....the view affinity matrix as input and permutes its rows and columns to generate a clustered view affinity matrix. The clusters are formed in such a way that views indexes with larger affinity values are collected together. Following the clustering, an earlierPartition algorithm described in [11] and [12] accepts the clustered affinity matrix and the view index usage matrix to produce fragments of the views indexes. The Partition algorithm finds sets of views indexes mostly accessed by distinct sets of applications. The Partition algorithm aims at finding the best point along the diagonal ....

....in Section 3. ffl Step 4: Compute the view index affinity matrix (VAM) as discussed in section 3. ffl Step 5: With view index affinity matrix, generate a clustered view index affinity matrix by running the Bond Energy algorithm (BEA 6 ) 9] 12] ffl Step 6: Run the Partition algorithm [11], 12] on the clustered affinity matrix to generate non overlapping fragments of views indexes 7 . ffl Step 7: Perform fragment cleaning to drop every view not accessed by a query and to drop which of a parent and child view pair has a higher benefit. ffl Step 8: Select fragment with higher ....

S.B. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical Partitioning Algorithms for Database Design. ACM Transactions on Database Systems, 9(4), 1984.

....basically brings up row 4 of the same matrix to row position 1 and the same rearrangement is undertaken for all rows. The clusters are formed in such a way that views indexes with larger affinity values are collected together. Following the clustering, an earlier Partition algorithm described in [8] and [9] accepts the clustered affinity matrix and the view index usage matrix to produce fragments of the views indexes. The Partition algorithm finds sets of views indexes mostly accessed by distinct sets of applications. The Partition algorithm aims at finding the best point along the diagonal ....

S.B. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical Partitioning Algorithms for Database Design. ACM Transactions on Database Systems, 9(4), 1984.

.... advantages of the proposed fragmentation and allocation scheme include maximal locality of query evaluation and minimization of communication cost in a distributed system, in addition to the desirable properties of (vertical) fragmentation and rule allocation as discussed in the literature [11, 12]. We also formulate the mathematical interpretation of the proposed vertical fragmentation and allocation algorithms. Keywords: rules, fragmentation, allocation, replication, deductive databases, distributed systems 1 Introduction Deductive database systems enhance the expressive power of ....

.... and unnecessary data transfer, increases the level of concurrency and therefore the system throughput in a distributed database system [6, 14] Vertical fragmentation further enhances the performance of database transactions by closely matching fragments for the requirements of transactions [12]. Our design goals of fragmentation and allocation of rule and data aim to maximize concurrent rule execution, reduce replication of rules and data, minimize communication cost during query evaluation, and decrease the query response time. Different approaches for (vertical) fragmentation and rule ....

[Article contains additional citation context not shown here]

S. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical Partitioning Algorithms for Database Design. ACM Transactions on Database Systems, 9(4):680--710, December 1984.

....features such as encapsulation, inheritance, class composition hierarchy. These features complicate the definition and representation of HP [8] The issues involved in distributed design for an object database system are presented in [14] and the authors suggest that the techniques used in RDBS [16] can be extended to object databases. However, they do not present an effective solution. In [12] and [13] the authors developed representation schemes for horizontal fragmentation, and presented a solution for supporting method transparency in OODBs. Ezeife et al. 7] presented a comprehensive ....

S.B. Navathe, S. Ceri, G. Wiederhold, and Dou J. Vertical partitioning algorithms for database design. ACM Transaction on Database Systems, 9(4):681--710, December 1984.

No context found.

S. B. Navathe, S. Ceri, G. Wiederhold, and J. Dou, "Vertical partitioning algorithms for database design," ACM Transactions on Database Systems, vol. 9, no. 4, pp. 680--710, 1984.

....Instead of precluding it, such architectural solutions complement data centric design. For example, clients can be allocated to each update server based on the potential e ectiveness of the resultant data centric grouping. One way to do this is to cluster clients based on interest anities as in [6]. However, because we assume a single update server, such work is outside the scope of this paper. 4. A HEURISTIC APPROACH TO GROUPING DESIGN Because of the complexity of the grouping problem (it can be modeled as an NP complete mathematical programming problem) we o er a heuristic algorithm. ....

S. B. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical partitioning algorithms for database design. ACM Transactions on Database Systems, 9(4), 1984.

....this process as employing data qualifications similar to the where clause in a SQL select command, making fine granularity replication possible. 13] Hence, given the transactions performed by each client, we can generate an initial fragmentation, and tune it with techniques described in [4, 23, 24]. As described in [27] fine grained units of updates (i.e. small fragments) improve the performance of client server synchronization, because they are more likely to contain little or no superfluous information. Although [27] suggests that tuple sized fragments are optimal for client ....

S. B. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical partitioning algorithms for database design. ACM Transactions on Database Systems, 9(4), 1984.

....This reduces the amount of irrelevant data accessed (thus, number of disk accesses) by the application. A fragmentation scheme partitions a relation either vertically (by clustering columns) or horizontally (by clustering rows) Algorithms have been developed to fragment a relation vertically [NCWD84, NR89, MCVN93, and others] or horizontally [CNP82, CNW83, and others] Allocation is the process of placing the fragments generated by the fragmentation scheme at the sites of the distributed database system so as to minimize the data transfer cost and the number of messages needed to process a ....

S. B. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical partitioning algorithms for database design. ACM Transactions on Database Systems, 9(4):680-- 710, 1984.

No context found.

S. B. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical partitioning algorithms for database design. ACM Transactions on Database Systems, 9(4):680--710, 1984.

No context found.

S. Navathe, S. Ceri, G. Wiederhold, and J. Dou, "Vertical Partitioning Algorithms for Database Design," ACM Transactions on Database Systems, Vol. 9 No. 4, pp. 680-710, 1984.

No context found.

Navathe, S. B., Ceri, S., Wiederhold, G. and Dou, J. (1984) Vertical partitioning algorithms for database design. ACM Trans. on Database Systems, 9, 680--710.

No context found.

S. B. Navathe, S. Ceri, G. Wiederhold, and J. Dou. Vertical partitioning algorithms for database design. ACM Transactions on Database Systems (TODS), 9(4):680--710, 1984.

First 50 documents

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