Document Type
Presentation
Publication Date
8-1-2014
Abstract
Hydro-model has become important tool for water resources management, with higher demand in simulation precision and speed of decision support, models designed for sectoral application becoming outmoded, and original mode that massive schemes are run sequentially cannot meet the real-time requirements, especially with the computation increase by finer discretization granularity and broader research range. Water management organizations are increasingly looking for new generation tools that allow integration across domains, and can provide extensible computing resources to assist their decision making processes. In response to this need, a hydro-modeling platform(HydroMP) based on cloud computing is designed and implemented, which can deployed in distributed HPC Cluster and center HPC Cluster use a resources balancer to manage load balancing. This platform integrates multi models and computing resources (i.e. blade computer) dynamically to assure models integrated in platform get extensible computing capacity. A server, hosting HydroMP Web Service and interfaces, is connected to the HPC Cluster and Internet constituting the gateway for registered users. Any terminal (i.e. decision making system) can reference library and Web service of HydroMP in their systems. Massive modeling schemes can be submitted by different users simultaneously, and terminal can get simulation results from HydroMP real-time. Some key approaches and techniques are utilized including: i) a standard model component wrapper communicating with platform by named pipe have developed. OpenMI-compliant model-components can be integrated to this wrapper; ii) API and Event-Handler interface provided by HPC Server, task scheduler and calculation management table is employed to dispatch computing resource, while controlling multiple concurrent scheme submitting; iii) Interface array(i.e. SchemesSubmit, StatusInquiry, GetResult) in the Web Service is supplied to make terminal communicate with platform; iv) Oracle database is used to manage massive model data, results and model-components. This paper describes the details of design and implementation, and gives a case presentation platform application.
Comments
Session R07, Web Services and Cloud Computing