Exploring a Self Organizing Multi Agent Approach for Service Discovery

2. Background

Web Services (WS) reliability is strongly dependent on the fault handling mechanisms of the communication protocols and on the messaging infrastructure mediating their interactions (Erradi, Maheshwari, & Tosic, 2006). Correct service delivery continuity defines web services reliability. This implies zero, or at worst, relatively few failures and rapid recovery time (Erradi, Maheshwari, & Tosic, 2006). WS are managed by different geographically distributed providers The unreliability of any of the constituent WS could lead to the failure or QoS degradation, even if other constituent seem to be reliable. A service management middleware is proposed in (Erradi, Maheshwari, & Tosic, 2006), it is based on some recovery policies (eg retry, skip, use). In (Ardissono L., Furnari, Goy, Petrone, & Segnan, 2006) a framework is proposed for WS orchestration, as environment of this latter could present some exceptions that are not able to identify precisely their causality. WS are used to diagnose exceptions in a more precise manner. In (Ardissono L., Furnari, Goy, Petrone, & Segnan, 2007) an interaction protocol for every cooperating web service WS_i is represented by an abstract process. The local view of WS_i on choreography is determined by associating its abstract process to the final WS. In this context, a WS monitor is responsible of checking the choreographed services by receiving their status messages during choreography. The problem is also posed when the WS monitor itself does not respond in time.