Article Preview
Top1. Introduction
As a result of the impressive proliferation of Internet-enabled devices, the growing adoption of SOA for implementing and deploying business applications on the Web and the cloud, organizations increasingly demand services that can meet their functional and quality-of-service (QoS) requirements. Therefore, service-providers need to negotiate Service Level Agreements (SLAs) with their service-consumers and honor them if they want to remain competitive. QoS refers to a set of qualities or characteristics of the service, such as response-time, availability, throughput, latency, security, reliability, and reputation. Several models and were proposed for representing QoS attributes of Web services. The most important ones are:
- 1.
Extension of the WSDL (Web Services Description Language) documents with QoS information (D’Ambrogio, 2006) (Kang, 2007).
- 2.
Extension of the UDDI (Universal Description Discovery and Integration) with QoS information (Lo, Cheng, Lin, & Chao, 2008).
- 3.
Utilization of a Broker (Yeom & Min, 2005) (Tao & Lin, 2005).
- 4.
Utilization of explicit QoS definition languages (Pal et al., 2000).
Service-providers have to adhere to their service level commitments if they want to remain competitive in a very changing and demanding business environment. Also, given the variety in consumers’ requirements, providers need to manage an increasing number of SLAs, all with potentially changing quality requirements. The SLA life cycle, as described in the SLA Management Handbook (TeleManagement Forum, 2004), includes six main phases: SLA Development, Negotiation and Sales, Implementation, Execution, Assessment, and Decommissioning. The most significant initiatives in standardizing the use of SLAs are (1) WSLA (Keller & Ludwig, 2003) by IBM, (2) the WS-Agreement specification (Andrieux et al.) by the Open Grid Effort, and (3) the SLAng specification Language (Lamanna, Skene, & Emmerich, 2003). Besides, many efforts have investigated the issue of SLA Negotiation that we discuss in the Related Work section (Dan, Davis, Kearney, & Keller, 2004) (Silaghi, Şerban, & Litan, 2012) (Hasselmeyer, Qu, Schubert, Koller, & Wieder, 2006) (Theilmann et al., 2010). This work shares with these efforts the common goal of providing support for automated SLA negotiation and management. Finding the right service-provider is a daunting task for consumers given the abundance and the variety of service offerings. Besides, collecting information about multiple service-providers is costly and time consuming.
Key aspects of the SLA-driven service provisioning, which are not addressed adequately by most SLA-based service provisioning systems, essentially concern multiple QoS attributes driven service selection, multi-issues SLA negotiation, and SLA-compliance monitoring. We propose in this work a framework that deals with these issues in an integrated way. The main components of the framework are: the Service Broker (SB) and the Monitoring Infrastructure (MI). SB mediates between service-consumers and service-providers, selects suitable service-providers that are capable of delivering required functionality and quality of service, and negotiates SLAs on behalf of service-consumers, and assesses compliance of service delivery with agreed upon SLA. Multiple rounds of negotiations are very often required to reach an agreement. In each round, SB and the selected service-provider bargain on several SLA parameters by trying to maximize their global utility functions. MI is in charge of observing service delivery, during SLA implementation, using measurements obtained from independent third party monitoring services.