Auction-based Admission Control for Continuous Queries in a Multi-Tenant DSMS
DocUID: 2012-004 Full Text:
PDF Author: Lory Al Moakar, Panos K. Chrysanthis, Christine Chung, Shenoda Guirguis, Alexandros Labrinidis, Panayiotis Neophytou, Kirk Pruhs
Abstract: The growing popularity of monitoring applications and "Big Data" analytics used by a variety of users will lead to a multi-tenant data stream management system. This paper deals with the problem of admission control of continuous queries, where the stream processing resources are sold to the end users. We employ variable pricing by means of auction-based mechanisms. The admission control auction mechanism determines which queries to admit, and how much to charge the user for each query in a way that maximizes system revenue. The admission mechanism is required to be strategyproof and sybil-immune, incentivizing users to use the system honestly. Specifically, we require that each user maximizes her payoff by bidding her true value of having a query run. We further consider the requirement that the mechanism be sybil-immune: that is, no user can increase her payoff by submitting queries that she does not value. Given the above requirements, the main challenges come from the difficulty of effectively utilizing shared processing of continuous queries. We design several payment mechanisms and experimentally evaluate them.
Keywords: Data Streams, Multi-Tenant, Admission Control, Game Theory, Auction Mechanisms, Strategyproof, Sybil-immune
Published In: International Journal of Next-Generation Computing
Volume: 3(3)Pages: 247-273
Year Published: 2012
Project: AQSIOS, STREAMS Subject Area: Data Streams
Publication Type: Journal Paper
Sponsor: NSF CAREER IIS-0746696, NSF IIS-0534531, NSF IIS-1050301