Utility computing provides a pay-as-you-go approach to information systems in which Application Providers (e.g., web sites) can better manage their costs by adding capacity in response to increased demands and shedding capacity when it is no longer needed [24]. This paper develops a framework for Application Providers using a cluster design to determine the number of servers that minimizes the sum of quality of service (QoS) costs resulting from service level penalties and server holding costs. The server characteristics considered are service rate, failure rates, repair rates, and costs. We further devise a simple approach for owners of computing utilities to select among servers with different characteristics and for server manufacturers to address design trade-offs between server characteristics. Our approach uses a performance and availability model that extends the M/M/m/K/M queueing system with considerations for failures and repairs. The model is validated with data from a product level testbed with an eCommerce workload. We develop a closed-form approximate solution for the optimal number of servers that works well for a wide range of system parameters. This result indicates that the optimal number of servers m* is the effective traffic intensity (the number of servers consumed by the workload with considerations for failures and repairs) with an adjustment for QoS costs and server holding costs. Last, we show that the Application Provider cost at m* is an increasing function of v, the effective server price performance (which considers failures and repairs). Thus, we propose that utility owners should consider v in their purchasing decisions, and manufacturers of servers should assess design trade-offs in terms of their impact on v.
By: Joseph Hellerstein, Kaan Katircioglu, Maheswaran Surendra
Published in: RC23325 in 2003
LIMITED DISTRIBUTION NOTICE:
This Research Report is available. This report has been submitted for publication outside of IBM and will probably be copyrighted if accepted for publication. It has been issued as a Research Report for early dissemination of its contents. In view of the transfer of copyright to the outside publisher, its distribution outside of IBM prior to publication should be limited to peer communications and specific requests. After outside publication, requests should be filled only by reprints or legally obtained copies of the article (e.g., payment of royalties). I have read and understand this notice and am a member of the scientific community outside or inside of IBM seeking a single copy only.
Questions about this service can be mailed to reports@us.ibm.com .