A performance model of speculative prefetching in distributed information systems

Abstract

Previous studies in speculative prefetching focus on building and evaluating access models for the purpose of access prediction. This paper investigates a complementary area which has been largely ignored, that of performance modelling. We use improvement in access time as the performance metric, for which we derive a formula in terms of resource parameters (time available and time required for prefetching) and speculative parameters (probabilities for next access). The performance maximisation problem is expressed as a stretch knapsack problem. We develop an algorithm to maximise the improvement in access time by solving the stretch knapsack problem, using theoretically proven apparatus to reduce the search space. Integration between speculative prefetching and caching is also investigated, albeit under the assumption of equal item sizes.

Performance modelling of speculative prefetching for compound requests in low bandwidth networks

“…We address the performance issue in [18] where we analyze the access improvement of a generic prefetcher. In [19], we develop integrated prefetch-cache algorithms to maximize the access improvement. The current paper is an extension of our earlier work.…”

“…In particular, we use a measure called access improvement, which is the improvement in the responsiveness of the system in real time units. In [19], we formulate the access improvement due to speculative prefetching for request to single item. In this paper, we generalize to allow for compound requests.…”

Analysis of speculative prefetching

“…The requested document will be randomly chosen out of an ordered set of documents . If a new request is generated during a speculative retrieval for another document, the speculative retrieval is immediately interrupted, whereas in [6] the speculative prefetch completes before the actually requested document is retrieved.…”

“…Previous research on algorithms and architectures for conventional caching [3] has been performed and the obtained results are a valuable information source for the investigation of prefetching. A model for speculative prefetching has been presented by Tuah et al in [6]. If this model is assumed, an optimum algorithm requires solving a stretch knapsack problem (SKP).…”

Effect of speculative prefetching on network load in distributed systems

“…In [11,12], performance of speculative prefetching is modelled and analysed, leading to an integrated approach to prefetching and cache replacement decisions. However, the analysis has not considered the effect of prefetching on network load.…”