QoS Aggregation in Web Service Compositions

Service Composition Quality Evaluation in SPICE Platform

“…Optimal selection of service components is a key issue for creating a service composition [ 31], but service description must provide more information that can be used to drive the developer during the composition design. Jaeger et al [15] proposed a mechanism to determine the overall Quality-ofService (QoS) of a composition by aggregating the quality attributes of the individual services: they identify abstract composition patterns, which represent basic structural elements of a composition like a sequence, a loop, or a parallel execution and they define aggregation functions for each quality attribute: this theoretical approach is used and partially implemented in this work to calculate aggregate values of some non-functional properties, as described in section 5. For more details on automatic service composition issues see [13] [23].…”

“…In particular, this figure shows the values of non-functional properties. The Aggregator can extract this graph from the SPATEL composition and it collapses the nodes in a each sequential path, calculating the aggregated values of the non-functional properties, using the above-mentioned aggregation functions [15]. After a first transformation where all sequential paths are collapsed it is time to collapse parallel nodes in a single one, then the graph is transformed in the one in figure 6; then the algorithm restarts collapsing sequential paths followed by parallel ones.…”

“…Moreover non-functional properties refer to the same definition in the above-mentioned Non-Functional Properties ontology, where their unit of measure is defined along with transformations among different units of measure. Given a SPATEL representation of a service composition, the Aggregator identifies abstract composition patterns (as defined in the previous work of Jaeger et al [15]), which represent basic structural elements of a composition, like a sequence, a loop, or a parallel execution. The aggregation of non-functional properties is based on an algorithm which recognizes composition patterns occurring in workflow and orchestration languages (like BPEL), it applies existing aggregation functions [15], and invokes the specific aggregator components to obtain aggregated values for these patterns.…”

QoS Explorer: A Tool for Exploring QoS in Composed Services

“…While copious work has been done recently in this area [14,2,19], much of it is predicated on the existence of a competitive market in compatible services, as well as mechanisms for automatic negotiation and entry into contracts with the providers of such services. Such a situation is not yet the case, and QoS Explorer addresses the real current situation of composition development and service contract negotiation still being a largely human-centered practice.…”

“…This facilitates design and instantiation of workflows to satisfy QoS constraints, as it enables the user to discover and focus effort on the aspects of a workflow which most affect their primary QoS concerns, thus improving efficiency of workflow development. Further, the underlying model we use is more sophisticated than those of similar recent work [14,2,19], and includes processing of entire statistical distributions and probabilistic states (instead of the simple numeric constants used elsewhere) to model such nonconstant variables as execution time. …”