Processing aggregate relational queries with hard time constraints

  • Authors:

  • Wen-Chi Hou;Gultekin Ozsoyoglu;Baldeo K. Taneja

  • Affiliations:

  • Case Western Reserve Univ., Cleveland, OH;Case Western Reserve Univ., Cleveland, OH;Case Western Reserve Univ., Cleveland, OH

  • Venue:
  • SIGMOD '89 Proceedings of the 1989 ACM SIGMOD international conference on Management of data
  • Year:
  • 1989

Quantified Score

Hi-index 0.01

Visualization

Abstract

We consider those database environments in which queries have strict timing constraints, and develop a time-constrained query evaluation methodology. For aggregate relational algebra queries, we describe a time constrained query evaluation algorithm. The algorithm, which is implemented in our prototype DBMS, iteratively samples from input relations, and evaluates the associated estimators developed in our previous work, until a stopping criterion (e.g., a time quota or a desired error range) is satisfied.To determine sample sizes at each stage of the iteration (so that the time quota will not be overspent) we need to have (a) accurate sample selectivity estimations of the RA operators in the query, (b) precise time cost formulas, and (c) good time-control strategies. To estimate the sample selectivities of RA operators, we use a runtime sample selectivity estimation and improvement approach which is flexible. For query time estimations, we use time-cost formulas which are adaptive and precise. To use the time quota efficiently, we propose statistical and heuristic time-control strategies to control the risk of overspending the time quota. Preliminary evaluation of the implemented prototype is also presented.