We present a new approach to predicting page accesses to facilitate prefetching persistent objects in a client/server computing environment. The relationships between persistent objects are modelled by a discrete-time Markov Chain, which allowed us to use a method called hitting times to compute the page access probability and the mean time to access a page. If the probability of a page is higher than a threshold defined by cost/benefit parameters then the page is a candidate for prefetching. To determine the prefetching threshold we consider various cost parameters to compare the benefit of a correct prefetch with the cost of an incorrect prefetch. In addition, we compute the best possible time to start the prefetch operation. We incorporated this prefetching algorithm into the EXODUS storage manager, and used the timing results in a simulation. Keywords: prefetching, persistent object stores, object-oriented databases, discrete-time Markov Chains, hitting times and hitting probabil...

I present the problem of client object cache lockup in a distributed object-oriented database system where storage is recovered by garbage collection. If garbage collection is augmented with the ability to discard reachable unmodified versions of persistent objects, cache lockup is greatly reduced. I outline some remaining questions about this technique for managing an object cache. 1 Introduction In a distributed object-oriented database system, server machines store persistent objects shared by applications running on client machines. When an application invokes an operation on a persistent object, that operation must run on either the server or client. I consider the case where objects are moved or copied to the client for at least the duration of the client transaction. A number of existing object-oriented databases work partially or entirely in this mode of executing operations on the client machine: examples are O2 [1], GemStone [2, 9], and Orion [7]. 2 Complicating factors The...

this paper, we examine the way in which prefetching can exploit parallelism. Prefetching has been studied before in many areas of computing such as operating systems, microprocessor design, compiler construction, the world wide web and databases. Although prefetching has been studied for a long time, the problem prefetching is facing has changed over the years as a result of technology advances. Fig. 1 shows the performance/price development of semiconductor memories and magnetic disks. There is a two-order of magnitude gap in access time between memory and disks. Memory access is faster and the rate of improvement is also higher. The prices of memory are falling which makes database N. Knafla / An Adaptable Multithreaded Prefetching Technique 3 caches cheaper and buffer replacement less of a problem. Prices of disks are also falling dramatically which is good for cheap secondary storage, allowing increased use of RAIDs. CPU performance improves at an even higher rate than memory (fig. 2). CPU is doubling its performance every 18 months whereas disk retrieval time only improves about 5% per year. Memory access time improves at about 10 to 12% per year. Client workstations will become powerful multi-processors machines with high speed CPUs, most of which will tend to be idle most of the time. 1 10 100 1000 1980 1985 1990 1995

In this report we compare the performance of a prefetching page server system with a prefetching object server system. We simulate the object access pattern by assigning transition probabilities to the object relationships. According to the transition probabilities we compute the access probability of pages and objects. We designed several prefetching techniques for a prefetching page server and a prefetching object server. We compared the performance of the prefetching techniques in a simulation. Keywords: prefetching, object-oriented databases, high performance object stores, persistence, client/server computing, disk storage management, page server, object server 1 Introduction There have been many efforts to integrate prefetching techniques into databases [6, 14, 23, 7, 9, 17, 11, 15, 16]. Object-oriented database management systems (OODBMSs) and prefetching techniques can be classified according to the unit of transfer between client and server. Some OODBMSs transfer pages to ...