Elastic Data and Computing Service for Real-Time Snow Assimilation
A GMU proposal submitted to NASA-ROSES-2012 A40:CMAC: Computational Modeling Algorithms and Cyberinfrastructure (NNH12ZDA001N-CMAC), $560,031, 3 mo/yr, 10/01/2012-09/30/2014.
NASA’s Computational Modeling Algorithms and Cyberinfrastructure (CMAC) Program invests in technology developments to reduce the risk and cost of evolving NASA information systems to support future Earth modeling efforts, including the integration of observational data into the model development, operations, and validation. A number of promising technologies have been demonstrated over the past five years which can be applied to the problems of correlating observational data for use in the models. The availability of large, virtualized pools of cloud computing resources raises the possibility of a new compute paradigm for scientific research.
Scientific cloud computing is emerging as a promising paradigm capable of providing an elastic, flexible, dynamic, resilient and cost effective infrastructure for evolving information systems to support Earth modeling environments. However, only certain scientific applications may optimally benefit from cloud deployment, and even these have special requirements that require tailored cloud solutions. Scientific cloud computing challenges include communication and I/O (input/output), creating cloud-aware scientific applications, scalability and reliability, programming and system administration to support science production, and science-specific security practices and policies. Therefore, we propose to study and optimize cloud computing for reducing the risk and cost of evolving NASA information systems to support future Earth modeling efforts. Specifically, we will evolve NASA’s Land Information System (LIS) modeling system and the Land Atmosphere Near real-time Capability for EOS (LANCE) into cloud-aware services, and understand and optimize the elasticity and performance of scientific cloud computing for the model-data interface through a series of test cases. The test cases will focus on the near-real time assimilation of high-resolution LANCE snow information into the LIS, and will address (a) measuring and optimizing I/O issues, (b) scalability of the model-data services, (c) adapting to real-time reliability issues, and (d) evaluate scientific performance under different cloud services.
Milestones: We will establish a NASA-GMU collaboration team deeply connected with relevant communities to incrementally concept, test, and prototype an elastic scientific model-data cloud computing environment.
1. A team will be formed and connected to communities. 2. Scientific, LIS model application, LANCE data services, and collaboration requirements for the elastic cloud computing environment and the 4 test cases will be analyzed and documented. 3. Detailed design of the scientific cloud environment, LIS model, and LANCE data service requirements will be setup.
4. Data and modeling services will be migrated to the cloud environment and made cloud aware. 5. Fourfour test cases will be implemented on the GMU cloud computing to understand and optimize the elasticity and performance of scientific cloud computing for the model-data interface, including: a. Measuring and optimizing I/O issues to enable optimum data-model interaction.
b. Exercising the scalability of the cloud enabled model-data services.
c. Adapting to real-time reliability issues. d. Evaluating scientific performance under 4 different cloud services. 6. Analyze the test case performance on the cloud with respect to user requirements and needs.