Towards Autonomic Service Provisioning Systems

Abstract

This paper discusses our experience in building SPIRE, an autonomic system for service provision. The architecture consists of a set of hosted Web Services subject to QoS constraints, and a certain number of servers used to run session-based traffic. Customers pay for having their jobs run, but require in turn certain quality guarantees: there are different SLAs specifying charges for running jobs and penalties for failing to meet promised performance metrics. The system is driven by an utility function, aiming at optimizing the average earned revenue per unit time. Demand and performance statistics are collected, while traffic parameters are estimated in order to make dynamic decisions concerning server allocation and admission control. Different utility functions are introduced and a number of experiments aiming at testing their performance are discussed. Results show that revenues can be dramatically improved by imposing suitable conditions for accepting incoming traffic; the proposed system performs well under different traffic settings, and it successfully adapts to changes in the operating environment.Comment: 11 pages, 9 Figures, http://www.wipo.int/pctdb/en/wo.jsp?WO=201002636

A Web Services Container Supporting QoS Hierarchical Control with Multiple Measurements for Utilization

“…Because all of the three hierarchical control strategies in [1], [3] and [5] are based on resource allocation and measure the utilization of web services container using the sum of the utilization for every service, those three hierarchical control strategies meet the contracts defined in Equation 3 and Equation 4.…”

“…Almeida et al [5] and L. Zhang et al [7] considered achieving QoS hierarchical control strategies measuring utilization in resource consumption aspect. J. Almeida et al [4] provided a framework focusing on short-term resource allocation and admission control to minimize resource usage costs. The QoS hierarchical control strategy in this model is based on the assumption that the distribution of the service request load should meet M/G/1.…”

“…Meanwhile, V. Kanodia et al [2] proposed QoS hierarchical control strategy based on admission control where value of control parameters are configured according to service response time of SLA and service average response time, based on the assumption that the service request load distribution meets M/M/1. [3,4] proposed a QoS hierarchical control strategy measuring utilization based on resource allocation in a commercial environment. They propose a financial model where service consumers pay for usage of services with QoS requirements and service provider offers services as well as guarantees of the QoS of services in return.…”

“…One reason is that the QoS hierarchical control with one measurement for utilization has limitations. R. Levy et al [1] and V. Kanodia et al [2] realized QoS hierarchical control using service response time to measure the utilization, which can significantly improve the performance of the container, but cannot separate the services with the same process time; M. Mazzucco et al [3,4] achieved QoS hierarchical control which calculates utilization using service revenue and penalty. Though the strategy can separate services flexibly, it has strong randomness because the revenue and penalty of the service are set manually; J. Almeida et al [5] implemented QoS hierarchical control from resource perspective, which can improve performance as well as saving resource.…”

Exploratory Performance Testing Using Reinforcement Learning

“…For instance, about 40% of the customers will abandon a web application if the response time is greater than 3 seconds [2]. Furthermore, Amazon, a leading online retailer, reports that 100 milliseconds (ms) additional delay in response time could cost them 1% drop in sales [3]. Therefore, ensuring the reliability and efficiency of a software system is imperative for such companies and the overall success of the software projects.…”