Optimal application performance on a Distributed Object Based System (DOBS) requires class fragmentation and the development of allocation schemes to place fragments at distributed sites so data transfer is minimized. Fragmentation enhances application performance by reducing the amount of irrelevant data accessed and the amount of data transferred unnecessarily between distributed sites. Algorithms for effecting horizontal and vertical fragmentation of relations exist, but fragmentation techniques for class objects in a distributed object based system are yet to appear in the literature. This paper first reviews a taxonomy of the fragmentation problem in a distributed object base. The paper then contributes by presenting a comprehensive set of algorithms for horizontally fragmenting the four realizable class models on the taxonomy. The fundamental approach is top--down, where the entity of fragmentation is the class object. Our approach consists of first generating primary horizontal fragments of a class based on only applications accessing this class, and secondly generating derived horizontal fragments of the class arising from primary fragments of its subclasses, its complex attributes (contained classes), and/or its complex methods classes.

Processing costs in distributed environments is most often dominated by the network communications required for interprocess communication. It is well-known from distributed relational database design research that careful placement of data "near" the users or processors where it is used is mandatory or system performance will suffer greatly. Data placement in relational database systems is comparatively simple because the data is flat, structured, and passive. Objects are characterized by an inheritance hierarchy (other hierarchies could also be considered including, class composition and execution), unstructured (possibly dynamic data), and contain a behavioral component that defines how the "data" is accessed by encapsulating it within the object per se. Algorithms currently exist for fragmenting relations, but the fragmentation and allocation of objects is still a relatively untouched field of study. Similar to relations, objects can be fragmented both horizontally and ve...

We present the design of a new transaction mechanism for an object-oriented database system called Thor. We also describe a mechanism that allows objects to migrate from one server to another. Our

Object-oriented, persistent programming languages offer a simple model for the programmer to write applications that share data, even among heterogeneous systems. However, poor performance limits their general acceptance. I present two independent but complementary techniques to speed up object operations in Theta, a object-oriented, persistent programming language that is used within the object-oriented database Thor. I examine the overhead that such a system imposes on operations such as method calls and field accesses, and show how these operations can be made fast without sacrificing the flexibility and extensibility of Theta. First, I describe a novel object layout that allows method dispatch as fast or faster than C++ implementations, while keeping the fast field accesses associated with nonobject -oriented systems. This layout derives from insights about the separation of subtyping and inheritance. Second, I show how customization can be applied to Theta to avoid many method dis...

Thor is a new object-oriented database system being developed at MIT. It allows applications written in different programming languages, and possibly running on heterogeneous machines and operating systems, to share objects conveniently. Our goal is to provide safe sharing of objects with higher-level semantics than is typical for today's file systems and databases, while still providing good performance. This paper describes the interface of Thor and also discusses some of the implementation techniques we are using to achieve our performance goal. 1 Introduction Most organizations run their applications on heterogeneous equipment. Not only do these computers differ in their hardware, but they frequently run different software, e.g., different operating systems, as well. In addition, applications may be implemented in different programming languages and more than one language might even be used within a single application. Nevertheless, applications may need to share information in sp...

Many researchers have demonstrated the importance of entity fragmentation in distributed relational database design. Database design will be essential in the "next-generation" engineering design environment that exploits object-oriented technologies. Fragmentation enhances application performance by reducing the amount of irrelevant data accessed and the amount of data transferred unnecessarily between distributed sites. Algorithms for effecting horizontal and vertical fragmentation of relations exist, but fragmentation techniques for class objects in a distributed object based system have not appeared in the literature. This paper first presents a taxonomy of the fragmentation problem in a distributed object based system capable of supporting systems engineering applications. Detailed horizontal fragmentation algorithms are then presented for one of these class models using a top--down approach where the entity of fragmentation is the class object. The algorithms described i...

this paper I propose three disk management strategies for object-oriented databases. These strategies are based on earlier work on file-systems. They differ from this earlier work in their support for a large number of small objects and non-sequential access patterns.

Object-oriented, persistent programming languages offer a simple model for the programmer to write applications that share data, even among heterogeneous systems. However, poor performance limits their general acceptance.

The increasing importance of distributed information systems calls for a better understanding of the nature of their use, since performance improvement and scalability of such systems depend heavily on this understanding. Over the last few years, the World-Wide Web (WWW or Web) has emerged as a unifying infrastructure for such large-scale information systems. The analysis of WWW usage can be done both at the client and/or at the server side. Similarly, performance enhancements are possible both at the client and/or at the server side. This thesis presents a detailed analysis of WWW usage and the performance enhancements that are possible to achieve as a result of such an analysis. On the client side, this dissertation presents pre-fetching techniques that alleviate the problem of long response times, by trading in network bandwidth for timeliness. On the server side, the dissertation shows replica allocation techniques that alleviate the problems of server load balancing and network b...