Improvement of Simulation Platform for providing reliable and easy use model simulation environment

We are developing a web site to offer a virtual machine (VM) for simulating mathematical model. Visiters can execute model programs or scripts without installing any software and tools on own computer. In the Simulation Platform (http://sim.neuroinf.jp/), VM is assigned and simulation is executed automatically on it. Each VM is run with Linux operation system, and installed basic software for computational neuroscience, including developer tools such as C++ compilers and libraries, popular neural simulators such as GENESIS, NEURON, and NEST, and scientific application software such as Gnuplot, R, and Octave. In this moment, our site is providing more than 250 contents, which are registered on the ModelDB and NIJC J-node platforms. The execution of contents is very simple, just selecting your choice and clicking the arrow mark on the screen shot of it. In the pre-release platform site, we have used our original software to manage VMs. It is work well for limited number of users and basic use, but it is considered to innovate more general VM technology for public release requested more stable and expandable operation.
In this study, we have developed our new VM system based on the IaaS (Infrastructure as a Service) technology. We examined various open-source IaaS frameworks for our backend system, according to the stability, usability, implementation and activity of the development. As a result, we judged that OpenNebula (http://opennebula.org/) was suitable for our purpose. OpenNebula has been actively developed for more than 4 years and is robust enough for large-scale public cloud systems. The system is composed of a small and efficient core written in C and a rich and friendly ecosystem written in Ruby that wraps the core. OpenNebula comes with a graphical user interface for system administration, which allows an administrator to add/delete hosts and VMs, monitor the system status, and manage users from a web browser (Fig.1).
Owing to the new system, it is not only enhanced reliability for many user accesses, but also scalability of our platform. It could be realized new functions, such as offering various VMs, which run on non-Linux operating system and different version of software. Based on the new VM, we start our service publically with runnable contents and their documents from this year. We will increase the volume and also type of contents, for example, 2D and 3D brain image viewing function for large-scale image data in the NIJC platforms, such as Invertebrate brain and Neuro-Imaging platforms.