Design And Implementation Of Web System For Hpgc Based On Django

With the emerging of more and more channels for Geo-spatial data acquisition, the complexity of Geo-spatial data processing is also becoming higher. Under this circumstance, High Performance Computing(HPC) plays an important role in these Geospatial data processing and applications due to its powerful computing capacity. As a result, a new research direction, High Performance Geo-Computation(HPGC), appeared and grows rapidly.Along with the continuous development of computer hardware technology, there are many new HPC platforms, such as GPU and Intel? Xeon Phi(Intel? Many Integrated Core Architecture, also called MIC). This greatly stimulates the research in both HPC and HPGC.However, there still exist many problems and barriers when users start using HPC devices/platforms. For instance, the operating system of most of HPC platforms is either Linux or UNIX, which are mainly operated with command lines. The configuration is quite complex. Users on those systems often need to develop or rephrase their Geo-spatial processing algorithms when they switch from non-Linux/Unix systems. The transition itself usually costs users lots of time. The traditional HPC application platforms/systems lack of some personified services, such as offering the remote visit via Web, or querying the status of submitted jobs. Obviously, to users new to Linux/Unix system, these problems will inevitably slow them down in using the HPC applications. In addressing such problems, the authors developed an HPGC application Web system based on Django framework. The main contributions focus on the followings.(1) Construct an HPGC application Web service system based on advanced computer technologies, such as RabbitMQ, Celery, and JQuery, under the Linux environment and Django development framework. Through the Web Portal of the developed system, a user can manipulate and manage the HPC platforms via the portal online. This capability can not only simplify the operations of the computing platform, but also facilitate the user to take advantage various HPGC algorithm services provided on the platform.(2) Design a unified algorithm encapsulation template that can facilitate the deployment of a user's Geo-spatial processing algorithm on the HPC platform. With the template, the system administrator can use sample operations to finish the algorithm's encapsulation. This capability greatly reduces the maintenance burden and enable service provides to offer more Geo-spatial algorithm services.(3) The HPGC application Web service system was designed to offer some convenient personified services to enable quick feedbacks to users. For example, the account management service can not only provide the e-mail message notification, but also invoke the SMS(Short Message Services) notification to keep users updated about the status of the submitted job.The HPGC application Web system can provide different algorithm services based on different HPC resources with a humanized service characteristic. In addition, the system is stable and easy to maintain.