چكيده به لاتين
Every state change of an event-driven system is reported by an event. Such systems can react to changes in a timely manner by using real-time processing of these events and deriving higher-level information. Real-time processing of events, detection of event patterns from events, and derivation of high-level information is called complex event processing (CEP). Patterns are defined by rules that are used by complex event processing engines. With increases in the deployments of event-driven systems, event generation rates are increased too, and systems need to detect various types of event patterns. Consequently, event-driven systems need scalable CEP systems capable of processing high rates of events using high numbers of rules. Vertical scaling is inadequate for applications of CEP in large-scale domains that require more computing power than available on a single machine with limited vertical scaling capabilities. Therefore, there is a need for the provision of mechanisms to scale CEP engines horizontally in support of large-scale event-driven systems. This thesis proposes SCF, a framework for horizontal scalability of complex event processing via event partitioning and rule partitioning. SCF consists of methodology, mathematical model, architecture and development environment. In addition, this thesis proposes a set of mechanisms under SCF using event partitioning, rule decomposition, and rule partitioning. In comparison with notable related works, the proposed mechanisms increased throughput nearly linearly when the number of CEP computing nodes increased. In addition, the proposed mechanisms distributed the processing load more balanced and had fewer false event detection rate.