Scheduling Update and Query Transactions under Quality Contracts in Web-Databases
DocUID: 2006-013 Full Text: PDFAuthor: Huiming Qu, Alexandros Labrinidis
Abstract: In modern web-database systems, users typically perform readonly queries, whereas all data updates are performed in the background, concurrently with queries. In this paper, we present the concept of Quality Contracts which allows individual users to express their preferences by assigning "profit" values to their expected Quality of Service (QoS) and Quality of Data (QoD). We propose an adaptive algorithm, called QUTS, to maximize the total profit from submitted Quality Contracts, which essentially optimizes the overall user satisfaction. QUTS address the problem of prioritizing the scheduling of updates (crucial to QoD) over queries (crucial to both QoS and QoD) using a two-level scheduling scheme that dynamically allocates CPU resources to updates and queries. We present the results of an extensive experimental study using real data (taken from a stock information web site), where we show that QUTS performs better than baseline algorithms under the entire spectrum of quality contracts; QUTS adapts fast to changing workloads and QUTS' sensitivity to its two parameters is negligible.
Published In: Proc. of the 5th Hellenic Data Management Symposium
Place Published: Thessaloniki, Greece
Year Published: 2006
Project: UserCentric Subject Area: Web Databases
Publication Type: Invited Paper
Sponsor: NSF ITR ANI-0325353