Abstract not found

Maintenance tends to degrade the structure of software, ultimately making maintenance more costly. At times, then, it is worthwhile to manipulate the structure of a system to make changes easier. However, it is shown that manual restructuring is an error-prone and expensive activity. By separating structural manipulations from other maintenance activities, the semantics of a system can be held constant by a tool, assuring that no errors are introduced by restructuring. To allow the maintenance team to focus on the aspects of restructuring and maintenance requiring human judgment, a transformation-based tool can be provided---based on a model that exploits preserving data flow-dependence and control flow-dependence---to automate the repetitive, errorprone, and computationally demanding aspects of re...

Using a trace of address references, we compared the efficiency of several different hashing functions, such as cyclic redundancy checking (CRC) polynomials, Fletcher checksum, folding of address octets using the exclusive-or operation, and bit extraction from the address. Guidelines are provided for determining the size of hash mask required to achieve a specified level of performance. 1 INTRODUCTION The trend toward networks becoming larger and faster, addresses becoming larger, has impelled a need to explore alternatives for fast address recognition. This problem is actually a special case of the general problem of searching through a large data base and finding the information associated with a given key. For example, Datalink adapters on local area networks (LAN) need to recognize the multicast destination addresses of frames on the LAN. Bridges, used to interconnect two or more LANs, have to recognize the destination addresses of every frame and decide quickly whether to receive...

In this paper, three techniques to implement logical OIDs are thoroughly evaluated: hashing, B-trees and a technique called direct mapping. Among these three techniques, direct mapping is the most robust; it induces at most one page fault to map an OID, and it scales very well to large, rapidly growing databases. Furthermore, the clustering of handles that are used to map logical OIDs is studied. In particular, the performance of B-trees and direct mapping can improve significantly if the handles of objects that are frequently accessed by the same methods are clustered. For direct mapping, two placement policies are compared: linear and matrix clustering. 1 Introduction The full support of object identity is one of the most important features of object-oriented database systems [KM94]. To improve referential integrity, an object base system allocates an object identifier (OID) to every object at the time the object is created. The OID is used to identify the object uniquely and to im...

UNIX support of disk oriented hashing was originally provided by dbm [ATT79] and subsequently improved upon in ndbm [BSD86]. In AT&T System V, in-memory hashed storage and access support was added in the hsearch library routines [ATT85]. The result is a system with two incompatible hashing schemes, each with its own set of shortcomings. This paper presents the design and performance characteristics of a new hashing package providing a superset of the functionality provided by dbm and hsearch. The new package uses linear hashing to provide efficient support of both memory based and disk based hash tables with performance superior to both dbm and hsearch under most conditions. Introduction Current UNIX systems offer two forms of hashed data access. Dbm and its derivatives provide keyed access to disk resident data while hsearch provides access for memory resident data. These two access methods are incompatible in that memory resident hash tables may not be stored on disk and disk resid...

This paper presents a distributed file organization for record-structured, disk-resident files with key-based exact-match access. The file is organized into buckets that are spread across multiple servers, where a server may hold multiple buckets. Client requests are serviced by mapping keys onto buckets and looking up the corresponding server in an address table. Dynamic growth in terms of file size and access load is supported by bucket splits and bucket migration onto other existing or newly acquired servers. The significant and challenging problem addressed here is how to achieve scalability so that both the file size and the client throughput can be scaled up by linearly increasing the number of servers and dynamically redistributing data. Unlike previous work with similar objectives, our data redistribution considers explicitly the cost/performance ratio of the system by aiming to minimize the number of servers that are acquired to provide the required performance. A new server i...

LH*lh is a new data structure for scalable high-performance hash les on the increasingly popular switched multicomputers, i.e., MIMD multiprocessor machines with distributed RAM memory and without shared memory. An LH*lh le scales up gracefully over available processors and the distributed memory, easily reaching Gbytes. Address calculus does not require any centralized component that could lead to a hot- spot. Access times to the le can be under a millisecond and the le can be used in parallel by several client processors. We showthe LH*lh design, and report on the performance analysis. This includes experiments on the Parsytec GC/PowerPlus multicomputer with up to 128 Power PCs and 32 MB of distributed RAM per node. We prove the e ciency of the method and justify various algorithmic choices that were made. LH*lh opens a new perspective for high-performance applications, especially for the database management of new types of data and in real-time environments.

Modern software systems for application areas like software engineering, CAD, or office automation are usually highly interactive and deal with rather complex object structures. For the realization of these systems a nonstandard database system is needed which is able to efficiently handle different types of coarse- and fine-grained objects (like documents and paragraphs), hierarchical and non-hierarchical relations between objects (like composition-links and cross-references), and finally attributes of rather different size (like chapter numbers and bitmaps). Furthermore, this database system should support incremental computation of derived data, undo/redo of data modifications, error recovery from system crashes, and version control mechanisms. In this paper, we describe the underlying data model and the functionality of GRAS, a database system which has been designed according to the requirements mentioned above. Furthermore, we motivate our central design decisions concerning its ...

IN THIS thesis, a new methodology is presented which supports the efficient representation, indexing and retrieval of images by content. Images may be indexed and accessed based on spatial relationships between objects, properties of individual objects, and properties of object classes. In particular, images are first decomposed into groups of objects, called "image subsets", and are indexed by computing addresses to all such groups. All groups up to a predefined maximum size are considered. This methodology supports the efficient processing of queries by image example and avoids exhaustive searching through the entire image database.

We propose an architecture for implementing nested re- In this paper we propose an implementation, ANDA’ lational databases. In particular, we discuss the storage for the Nested Relational Data Model(NRJ3M). In parstructures, their organization and an access language titular, we diicuaa the storage structures, their orgafor specifying access plans. nization, and an access language for specifying access The featurw of our implementation are: plans. The motivation for our design comes from these observations: A notation for hierarchical tuple identification. One value-driven indexing structure (VALTREE) for the entire database. A main-memory based component (CACHE) for ma-nipulating hierarchical tuple-identifiers. A hashing scheme (RECLISTS) for fast access to data specified by tuple-identifiers. An access language based on the VALTREE, the RECLIST and the CACHE to define access plans for execution of queries. Pemusaon to copy without fee all ar put of this granted pvided that the oopies are not made or disrritutcd for direct commercial advantage, the VLDB cop&ht notice md the title of the public&m and its date s~~eer, and notice L given