A stochastic evaluation model for database organizations in data retrieval systems

Abstract

Experimental work in the valuation of large scale data retrieval systems has been scarce due to its difficulty and prohibitive cost. This paper discusses a simulation model of a data retrieval system which has the effect of significantly reducing the cost of experimentation and enabling research never attempted before. The model is designed to estimate the retrieval workload of alternative data retrieval systems. These data retrieval systems can be organized under several database organizations, including inverted list, threaded list, and cellular list organizations and hybrid combinations of these systems. Effectiveness of the methodology is demonstrated by using the model to study the effect of database organizations in data retrieval systems. In particular, the impact of query complexity is analyzed.

On estimating block accesses in database organizations

“…then there will be no reduction. We shall assume as Lowe [3] and Siler [4] did. that the number of tuples with the same A value is a random variable and is distributed according to Zipf's distribution [9].…”

An attribute based model for database access cost analysis

“…THEOREM 1. The expected response time for a query of complexity (d, c, p, q) is ~9 = S(L 5, P) + MS + 1, t, q) + Nt + 1, r, PP) + @r + 1, n, El, (13) where E = l~y,(q/N,)"-'Id.…”

“…Examples of single-attribute organization analysis include the VSAM model [6]. Previous models for multiattribute database organizations include the FOREM simulation model [lo], the analytic models developed by Lowe [8] and Martin [9], and the simulation models reported by Cardenas [2] and Siler [13], In each case, a given structure is analyzed and an analytic or simulation evaluation is performed. Since different assumptions and techniques are used in different analyses, it is dilhcult to compare results from individual models.…”

A Practical Approach to Selecting Record Access Paths

“…The answer is that problem and solution characteristics affect system performance in complex ways; their subtle relationships defy exact analysis. Though a noble quest for researchers [6,12,20,45,49,54], the search for optimal database designs seems to be an impossible task.…”