Cloud Computing Test Bed for NASA Earth Observation

Abstract:

In order to develop a deeper understanding of utilizing cloud computing technologies for using earth observation data processing a test bed was created to ease access to the technology. Users had expressed concerns about accruing large compute bills by accident while they are learning to use the technology.

The test bed is to support NASA efforts such as:

• Developing a Science Data Service platform to handle big earth data for supporting scalable time and space searches, on-the-fly climatologies, data extraction and data transformation such as data re-gridding.
• Multi-sensor climate data fusion where users can select, merge and cache variables from multiple sensors to compare data over multiple years.
• Facilitate rapid prototype efforts to provide an infrastructure so that new development efforts do not need to spend time and effort obtaining a platform. Once successful development is done the application could then scale to very large platform on larger or commercial clouds.

Goals of the test bed are:

• To provide a greater understanding of cloud computing so informed choices can be made on future efforts to handle the over 15 Petabytes of NASA earth science data.
• Provide an environment where a set of science tools can be developed and reused by multiple earth science disciplines.
• Develop a Platform as a Service (PaaS) capability for general earth science use.

This talk will present the lessons learned from building a community cloud for earth science data.