DOI: 10.5176/978-981-08-5837-7_152
Authors: Qianhui Liang & Michael N. Huhns
Abstract:
When Web services are moved into a cloud computing environment, each service can be scaled over more servers to handle increases in demand. However, there is still the problem that external applications using the services must move control and data information in and out of the cloud. If there is a large amount of data, it would be more efficient for the data to flow directly among the services in the cloud, rather than to and from the external controlling application. Currently, there is no way to specify this using SOAP + WSDL, or REST. To achieve the full benefits of cloud computing, we describe in this paper a technique for transitional storage called cloud data containers to be placed within the cloud to hold intermediate process data, such that the data traffic to and from the cloud can be reduced substantially. The assignment of the physical data containers to individual services relies on service dependency relationships modeled by a directed graph that we term a weighted service dependency graph. Dependent services share one or more data containers. We have designed a data container selection scheme to identify the containers best able to transmit data from one partner service to another. The validity and efficiency of our data container assignment and selection scheme are demonstrated by experiments.
Keywords: Cloud data containers, data traffic, intermediate process data, transitional storage, service flow, weighted service dependency graph
Updating...