Abstract: To provide reliable real-time cloud computing services for large-scale data access and massive marine information processing and by combining the idea of workflow and software-as-a-service (SaaS), a concept of software service flow and build a software service flow architectures system based on a cloud platform is proposed. In this system, a service flow engine is an underlying layer, which interacts with the Hadoop platform. When processing user requests, the engine runs a self-design algorithm which analyses and combines service flow, and is delivered to the underlying layers for execution. Moreover, for the sake of control service flow and manager resource, it also providers many transparent interfaces to the upper layers with representational state transfer (REST) style, thus reducing complexity of development and improving scalability of the system. Users can access the Web page, customize software services, and monitor the cloud platform on real-time. On this platform, large-scale data access, high concurrency, and high-density access are a normal status. By building an initial prototype system, the availability and efficiency of the SaaS-flow system structure is proved.

Key words: Hadoop, cloud computing, cloud platform, representational state transfer (REST) architecture, software-as-a-service (SaaS)-flow