Abstract

For e-commerce Web service applications, cache should be applicable especially to product information services. However, practical e-commerce use cases have limited cacheability: (1) a message may contain management components, which reduce reusability of cached messages; (2) content typically consists of composite objects, which makes it harder to maintain cache effectively. Our solution proposed in this paper is service decomposition that decomposes a web service into management services and composite objects. We introduce our cache model and specification based on composite objects. The specification includes dependency description that specifies how to maintain changes in composite objects when the source objects are updated. In order to implement cache management on this model, a large number of XPath query results over the source objects must be cached and effectively maintained. In this paper, we also introduce a high performance implementation scheme of cache management based on a hybrid XML filter that combines in-memory structure matching and RDBMS-based parameter matching. The filter takes advantage of parameterized query patterns within dependency description and achieves scalability.

Keywords: Web services; Cache; Edge computing

Article Outline

1. Introduction

2. Related work

3. Service decomposition architecture

3.1. Management service decomposition

3.2. Content decomposition

4. Object-based XML cache management

4.1. Composite object model

4.2. Cache specification

4.2.1. Object definition

4.2.2. Object list definition

4.2.3. Reference structure definition

4.2.4. Cache identity definition

4.2.5. Cache managed content definition

4.3. Dependency description

4.3.1. Trigger-based dependency description

4.3.2. Query-based dependency description

4.3.3. Use of dependency description

5. Implementation

6. Experiments

7. Conclusion

References