NASA's Movement Towards Cloud Computing


Published on

Chris C. Kemp, Chief Technology Officer for IT, NASA

Published in: Technology, Education
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide
  • The launch of the first artificial satellite by the then Soviet Union in 1957 marked the beginning of the utilization of space for science and commercial activity. During the Cold War, space was a prime area of competition between the Soviet Union and the U.S. 

In 1964 the first TV satellite was launched into a geostationary orbit to transmit the Olympic games from Tokyo. Later, Russian launch activities declined while other nations set up their own space programs. Thus, the number of objects in Earth orbit has increased steadily -- by 200 per year on average. 

The debris objects shown in the images are an artist's impression based on actual density data. However, the debris objects are shown at an exaggerated size to make them visible at the scale shown.

Image Credit: European Space Agency
  • Today there are over 100 active NASA missions, some of them bringing back a TB or two of data EVERY DAY. There are many challenges in handling this vast amount of data… (Note: This is an artist’s rendering of recent NASA missions)
  • NASA's Movement Towards Cloud Computing

    1. 1. Cloud Launch<br />The NASA Nebula Project<br />Chris C. Kemp<br />Chief Technology Officer for IT<br />1<br />
    2. 2. 2<br />National Aeronautics and Space Administration<br />
    3. 3. 3<br />National Aeronautics and Space Administration<br />
    4. 4. NASA’s IT Challenges<br />As a result of NASA’s diverse mission and distributed workforce (across centers, labs, and universities), we have a uniquely balkanized – and expensive - IT environment.<br />Typical Enterprise IT infrastructure is only utilized to 5-20% of its capacity1<br />NASA’s Supercomputers are also only available to the largest projects, with small projects often waiting many weeks in queue… many jobs do not require HPC.<br />Thousands of new computers are inefficiently procured and operated each year, continuing the cycle. This process often takes many months and costs more in overhead than the actual value of the computers.<br />1 Source: <br />4<br />National Aeronautics and Space Administration<br />
    5. 5. 5<br />National Aeronautics and Space Administration<br />
    6. 6. 6<br />
    7. 7. 7<br />
    8. 8. 8<br />
    9. 9. REST, CLI, Web Interfaces<br />9<br />
    10. 10. WISE in the Cloud<br />11<br />124<br />7<br />12<br />5<br />4<br />9<br />109<br /> 2<br /> 6<br />Nebula Users by Location<br />10<br />
    11. 11. Science 8X sky survey<br /><ul><li>Find the most luminous galaxies in the universe
    12. 12. Find the closest stars to the sun</li></ul>Primary Data Products<br /><ul><li>WISE Image Atlas</li></ul>10,464 calibrated FITS image sets (4 bands/set), 4kx4k pix @1.375”/pix<br />Formed by combining all single exposures covering Atlas Tile footprint<br />Pixel depth-of-coverage, uncertainty maps and metadata for each image<br /><ul><li>WISE Source Catalog</li></ul>J2000 positions, calibrated 4-band photometry, quality and value-added flags and parameters for over 200 million objects in the release area<br />Wide-Field Infrared Survey Explorer (WISE)<br />11<br />National Aeronautics and Space Administration<br />
    13. 13. WISE in the Cloud<br />#<br />Angular resolution processing. <br />Processing to resolve and measure source sizes.<br />1 1 1 100<br />1 1 1 2000<br />1 1 1 100,000<br />1 80 100 10<br />1 1 1 100<br />1 1 1 41,253<br />Processing to differentiate point-like emission from distributed emission.<br />Surveying for distributed star formations in our galaxy. <br />A few large regions can be processed to tremendous depth. <br />High resolution processing of the entire sky.<br />12<br />National Aeronautics and Space Administration<br />
    14. 14. <ul><li>An Ames researcher developed an algorithm to produce a useful result from standard ATC data.
    15. 15. Took 20 hrs on Mac laptop for each airport-day (or 83% of real time)
    16. 16. Nebula user asked to re-create researcher’s prototype and run a year’s worth of data.
    17. 17. Initial results: 10241 site-hours (about 1/7 year, first of six phases) in 17 compute hours
    18. 18. or ~85 times faster than the Mac Laptop
    19. 19. Using 4 tiny instances</li></ul>Air Traffic Control Algorithm<br />13<br />National Aeronautics and Space Administration<br />
    20. 20. Science<br /><ul><li>Numerical Weather Prediction
    21. 21. High-Res Satellite Imagery Processing</li></ul>Primary Data Products<br /><ul><li>Rapid Weather Modeling</li></ul>Shared resources for high resolution processing or large forecast domain.<br />Rapid responses to new events or research opportunities without impacting other resources.<br /><ul><li>Satellite Images</li></ul>Process and distribute high resolution images on demand using web-driven, scalable architecture capable of processing large datasets.<br />SERVIR and SPoRT<br />14<br />National Aeronautics and Space Administration<br />
    22. 22. User Notes<br />“After having access to Nebula for only a few days, I find that I am already able to accomplish more data-intensive calculations than I can do on any of the local servers we have here, and with no difficulties at all!”<br />“The system is easy to access and use, and offers a capability that I absolutely need occasionally but for which I could never justify the expense if it were for my needs alone.”<br />“With the recent addition of a large-RAM instance, I am now able to conduct calculations that could not be done on our project's large server farm.  Nebula has provided me with a tool for science data analysis that far surpasses anything that I could envision in a single-user context.  NASA Cloud computing may be the way forward for our data-intensive projects in the future, since only a NASA system could provide the necessary reliability and proprietary controls on our data.”<br />1 Source: <br />15<br />National Aeronautics and Space Administration<br />
    23. 23. Nebula’s Computing Engine, NOVA, is the foundation of OpenStack Compute. <br />OpenStack is to the datacenter what Linux is to the personal computer.<br />An open source cloud stack that enables any organization to build large scale private cloud services on commodity hardware. <br />Standard Apache 2.0 license<br />16<br />National Aeronautics and Space Administration<br />
    24. 24. 600 developers,100 companies,12 countries, 6 months <br />
    25. 25. Chris C. Kemp<br />Chief Technology Officer for IT<br /><br />@kemp<br />18<br />