Liangzhao Zeng, Boualem benatallah, et. al. QoS-aware middleware for Web Services Composition, tose, 2004
This is an extended version of www03. The content of the two papers is similar, but it is clearer and clearer in many places than that written in www03 (after all, it is written on tose ).Article).
1. This article introduces middleware agflow that supports QoS driven Web service composition. The main innovations include:
1) A set of web service quality model (Execution price, execution duration, reputation, successful execution rate, availability) is proposed, which can describe a single WS, and the aggregation method of QoS for composite services (based on execution path)
2) Two QoS-driven service selection methods are proposed: (s4.1) is based on local optimization. For a single WS, the QoS value is normalized and then evaluated using the saw method, this is relatively simple. (s4.2) is based on global planning. The main content is basically the same as that in www03. First, we will discuss a single execution path, then the multiple execution path, and finally propose to use integer programming (IP) method.
3) In addition, (s4.2.5) It is discussed that replanning is required when the WS (such as unavailable and QoS change) in the execution plan are being run or are to be run are changing. the replanning process is similar to the global planning at the beginning. The difference is that the QoS data of the executed WS is known at this time, and the running ws cannot be replaced.
2. Some assumption mentioned in this Article
1) "We assume that when the execution of Noncritical services is not successful, the task can be re-executed without delaying the whole composite service execution ."
2) "We assume that, when the noncritical service is unavailable, a service can be reselected without delaying the whole composite service execution ;"
3. Excerpt
(S4.2.3) "it can be proven that any arbitrary statechart can be transformed into an equivalent statechart in which the cycles do not cross the boundaries of conditional branches"
4. Miscellaneous
(S6 experimentation) not viewed