Abstract-The enabling of scientific experiments that are embarrassingly parallel, long running and data-intensive into a cloud-based execution environment is a desirable, though complex undertaking for many researchers. The management of such virtual environments is cumbersome and not necessarily within the core skill set for scientists and engineers.We present here Chiminey, a software platform that enables researchers to (i) run applications on both traditional highperformance computing and cloud-based computing infrastructures, (ii) handle failure during execution, (iii) curate and visualise execution outputs, (iv) share such data with collaborators or the public, and (v) search for publicly available data.Demo video: http://youtu.be/Twi-d2WT94AI. INTRODUCTION Researchers have been using computing resources such as desktops and supercomputers for running their experiments. In order to use such resources, researchers are expected to know how to set up their execution environment, run their experiments and collect and optionally share the output of their experiments. When executing computational experiments on a local desktop machine, performing these tasks may not be challenging. However, if an experiment's resource requirements exceed those of a single workstation, then computing environments such as cluster, grid or cloud can be considered.Cloud computing [3] presents a unique opportunity for users: it enables researchers to acquire very large numbers of computing and storage resources quickly. Moreover, researchers with relatively modest requirements for parallelisation of existing code may be able to avoid learning highperformance computing (HPC) infrastructure concepts. Researchers still need to learn how to work within a cloud-based environment, which itself presents its own challenges. They need to to create and set up virtual machines (VMs) in the cloud, collect the results of their experiments, and then release the VM resources. Furthermore, cloud-based environments are more prone to failure than HPC environments due to network and third-party software issues [13], and these environments expect researchers to handle such failures themselves.The rate of technological change and innovation for compute environments is ever increasing. When a new technology

INTRODUCTIONWe are developing the Virtual Experiences (Vx)Lab, a research and research training infrastructure and capability platform for global collaboration. VxLab comprises labs with visualisation capabilities, including underpinning networking to global points of presence, videoconferencing and high-performance computation, simulation and rendering, and sensors and actuators such as robotic instruments locally and in connected remote labs. VxLab has been used for industry projects in industrial automation, experimental research in cloud deployment, workshops and remote capability demonstrations, teaching advanced-level courses in games development, and student software engineering projects. Our goal is for resources to become a "catalyst" for IT-driven research results both within the university and with external industry partners. Use cases include: multi-disciplinary collaboration, prototyping and troubleshooting requiring multiple viewpoints and architectures, dashboards and decision support for global remote planning and operations, e.g. in automation, power, facilities or environmental management, and safe student access to remote or dangerous labs without "suiting up". VxLab was originally developed (as "VITELab") to support research collaboration within the Australia-India Research Centre for Automation Software Engineering (AICAUSE) [2,3,4], a partnership between ABB and RMIT University. ABB is a global leader in power and automation systems. In this talk we describe VxLab and its main use cases.The Global Operations Visualization (GOV) Lab supports local user collaboration while accessing and displaying standard applications and internet services through a high resolution 8m x 2m video display wall. Local services include video conference (H323 / skype) and streaming to/from local and remote sites. GOV Lab is designed for use with parallel rendering middleware such as SAGE middleware [1]. The figure below shows a user at the GOV Lab visualization wall.The Cyber-Physical Simulation (CS)Rack provides an experimental 640 core HP blade server, to support modeling, simulation and collaboration services. CSRack is publicly accessible and configured via OpenStack. Our intention is to explore the use of CSRack as a "cloudlet," locally-connected, low-latency and high-bandwidth, for use as a gateway to national e-research infrastructure such as NeCTAR. The Advanced Manufacturing Robotic Interoperation Test (AMRIT) Lab provides two ABB IRB120 industrial robot arms with three-fingered grippers, sensors and cameras. The VxLab Network is logically separate from the RMIT corporate network to create a sandbox environment. Dedicated private network links have been built to RMIT labs and to ABB's Melbourne and Bangalore sites (separate from the ABB corporate network). CHIMINEY The Chiminey system [5] is a cloud-based computing platform that enables scientists to perform complex computation on cloud-based and traditional high performance computing (HPC) facilities. Chiminey gives special importance to resource access ...

No abstract

No abstract