Dynamic Approach to the Construction of Progress Indicator for a Long Running SQL Queries

Widely used Data Warehousing (DW) and Business Intelligence (BI) technologies are mostly based on complex and long-running queries. It is very important for users to have information about query execution progress. Percent-done progress indicators are a technique that graphically shows query execution time (total and remaining) or degree of completion. In this paper, the authors propose a method that constructs model of a percent-done progress indicators based on adaptive approach. During the learning phase, the method analyzes the influence of averaged system state on the SQL query response time using data mining algorithms. The results of this phase are models representing query behavior under dynamic server workload. Built models are used during the production phase for the estimation of query execution time.