OpenStack NASA

OpenStack NASA

• 1.
  The Future ofCloud Computing at NASA Karen Petraska Service Executive, NASA Computer Services Office Raymond G. O’Brien CTO for IT, Ames Research Center April 20, 2012
• 2.
  Past, Present, Future Fromthe NASA Strategic Plan: Goal 6: Share NASA with the public, educators, and students to provide opportunities to participate in our mission, foster innovation and contribute to a strong National economy • 2009: NASA’s contribution to OpenStack was timely to the introduction of cloud computing to the industry • 2012: Industry has enthusiastically embraced OpenStack and an increasing number of commercial implementations of OpenStack clouds are now available • 2012 and Beyond: NASA shifts focus to becoming a wise and informed consumer of commercial cloud services
• 3.
  NASA CIO’s VisionFor Cloud Computing  The NASA CIO’s vision for cloud computing at NASA: » To have a good computing environment that addresses NASA’s computing requirements for all NASA people » Have an easy and seamless way to obtain services, leveraging economies of scale wherever possible » Innovate how we do security in the cloud; remove as much burden from the end users as possible » Leverage our buying power and unique requirements to influence industry where appropriate » Be agile and nimble to embrace and integrate new technologies that support our mission
• 4.
  Embracing the Technology The NASA community in general is starting understand the advantages of using the cloud model  NASA is evaluating its application portfolio and experimenting to understand the characteristics of applications that run well in the cloud  NASA is exploring options for and aspects of delivering enterprise cloud services within the NASA environment » Challenges: Governance, Security, Cost Recovery  Success stories in using commercial cloud » JPL BeAMartian and others » NASA Web Environment
• 5.
  NASA’s Computing Environment NASA requires many types of computing » Business and administrative (highly virtualized today) » Web sites and web applications » Modeling and simulation » Science data processing and analysis » Engineering analysis » Flight command, control, telemetry and flight operations  NASA collaborates with scientists and others all over the world » Universities, corporations, other US Government agencies, foreign space agencies  Data of interest to NASA resides in many locations depending on the collaborators » Often extremely large data sets » Science data archives of long term scientific interest
• 6.
  Actions to Achievethe Desired Future State  Current activities to support an enterprise cloud service » IaaS focus for now • Working to understand what platforms will be useful » FedRAMP: A&A for provider controls but what about consumer controls? » Common Cloud User and Management Interface » Acquisition Strategy » Best approach for user support • Decision trees for assessing cloud suitability • Managed environments
• 7.
  Things We’ve Learned
• 8.
  What Works Well(and What Doesn’t)  It is easier to birth new applications in the cloud than to migrate legacy applications » Legacy code that has not been ported recently can be very time consuming to move to cloud, and if it requires out of date or specialized compilers, it may simply not be worth the effort  Applications that are bursty or require many nodes for brief periods of time » Applications that run continuously may not be economical in the cloud  Compute servers that have more than 50% of wall clock time idle  Applications that need to be always available but not always running  “Embarrassingly parallel” computations (e.g., suitable for Hadoop processing) work well in the cloud » Applications that require significant inter-processor communication (e.g. climate models) will be substantially slowed
• 9.
  Economics  The learningcurve getting into cloud is steep: developers and systems administrators have to learn new paradigms and work in new ways  Second big investment is moving legacy code into the cloud  If the organization has excess capacity in existing data centers and existing owned computing infrastructure, it may make sense to fully utilize what you already own BEFORE paying for additional capacity in a cloud  Build a business case for each application to decide if a move to cloud is prudent: understand hidden and unanticipated costs. » Costs of getting data into and out of the cloud » Cost of computing cycles » Long term data storage costs » Licenses, IP addresses, etc.
• 10.
  Summary  NASA isembracing cloud computing  NASA is working to become a well educated customer of commercially available cloud services  We believe there are aspects of our business that can be improved through the use of cloud, potentially enabling more work within the same budget
• 11.
  Cloud’s Potential isBecoming More Evident Within NASA 2009 2012 April 20, 2012
• 12.
  Today: Lots ofNASA Piloting Activity  Different cloud providers  Mission and Enterprise workloads  Private and public cloud services  Becoming an informed consumer  Some already pursuing steps towards routine use 04/20/12
• 13.
  OpenStack: Creating Competition and Choice  NASA ultimately buys all its on-going services, support, and products  NASA will likely use most major commercial services in the future  Ditto for major private cloud products  As a cloud consumer, NASA wins from competition and choice 04/20/12
• 14.
  NASA’s Continued Community Involvement  NASA is very proud to have been part of the creation of OpenStack  Future participation will shift largely to involvement as a user  Very active Bay Area community makes this easy for Ames  Will strive to keep the original Nebula-based contributor authority active 04/20/12
• 15.
  Climbing up theStack within NASA  Lots of focus on IaaS layer right now within NASA, but…  Nebula testing feedback showed the high value certain groups place on what equates to platform services  Increasing popularity of SaaS with employees requires guidance and governance 04/20/12
• 16.
  Leveraging OpenStack’s Successfor Future NASA Policy Revision  Commercial Tech Transfer is a high priority for NASA  OpenStack is a shining example  Current NASA policy is being reviewed for possible revision to allow more NASA involvement in community SW development products 04/20/12
• 17.
  Product Positioning forFuture NASA Use  Federal Risk and Authorization Management Program (FedRAMP)  FedRAMP will be Key to Future Authorized NASA Cloud Use of Commercial Clouds  Cloud features that facilitate or address NIST 800-53 control implementation will be highly valued 04/20/12
• 18.
  NASA’s OpenStack Listfor Santa  Easy to install distributions  Many commercial service providers  Lots of value-add tool and support providers  Lots of priced enterprise- class customer support options  Tight alignment with FedRAMP and NIST security requirements  Niche features enabling HPC cloud  Continued rapid feature development 04/20/12
• 19.
  And Finally…  A much deserved thank you from NASA to: » The Nebula project team and all its sponsors and supporters » Rackspace Hosting » The entire OpenStack community  Together you have created something truly special that will benefit the industry and all consumers of cloud  Congratulations! 04/20/12