1 - Basic principles

Abstract

Database tuning is the activity of making a database application run more quickly. Tuning rests on a foundation of informed common sense. Five basic principles pervade performance considerations (1) think globally, (2) partitioning breaks bottlenecks, (3) start-up costs are high, (4) render unto server what is due unto server, and (5) be prepared for trade-offs. This chapter describes each of these principles and gives examples of its application. Effective tuning requires a proper identification of the problem and a minimalist intervention. This entails measuring the right quantities and coming to the right conclusions. Partitioning in database systems is a technique for reducing the load on a certain component of the system either by dividing the load over more resources or by spreading the load over time. Obtaining the best performance from a data-intensive system entails more than merely tuning the database management portion of that system. An important design question is the allocation of work between the database system (the server) and the application program (the client). Where a particular task should be allocated depends on three main factors: (1) the relative computing resources of client and server, (2) where the relevant information is located, and (3) whether the database task interacts with the screen.