Cloud Projects and Examples
This technial session invites speakers to discuss some of their current Cloud Computing projects. The focuses on refrence architecture, best practices, and pitfalls avoidance. Speakers include:
- Deirdre Byrne: Current NOAA cloud computing projects
- Long Pham: Giovanni on the cloud
- Mike Little: NASA Langley Cloud Practice
- Kai Liu: Cloud advisory tool
Long Pham NASA Goddard:
Cloud Computing with Giovanni (g4Cloud)
Objectives
Overall Appoach - Amazon and Eucoplyptus Cloud
The Giovanni Way - How does is work
Use-Case Scenarios
-workshop portal
-run Giovanni next to data
-mitigate peak load before major conferences
g4Cloud performance
-runs about 25% faster in the cloud than on local system
Cost comparison
-85K in the cloud vs 120K on local machines over 3-year period
GUI example / Use Example
Summary Information / Next Steps
-would like to scale g4Cloud on Amazon
-experiment with different datasets
Q:How scalable, can support I/O and bandwidth
A:Ran bandwidth test on local and Amazon instance, but not multiple instances
Q:Will you do more testing with Eucalyptus
A:Yes, we will
Other Questions....
| Live Demo |
______________________________________
Mike Little
Cloud Computing Experiences at NASA Langley
All work at Langley relates back to Nebula testing
Most code is custom / data reduction work
-Analysis shows that if you use more than 60% of the time, then you should probably own the computer
-No saving in system support cost because sys-admins just do other kind of support
-Experiment early,
-New projects are easier to move onto the cloud
Important Use Cases at Langley
-useful for startup projects
-foreign national visitors
-others
Cost model
-don't have a good one , constantly changing
Science community not always comfortable with new technologies
Business interface information
GeoCloud example - seems to be working well
Moving some of the data to the cloud
_____________________________________
Deirdre Byrne: Current NOAA cloud computing projects
NOAA Data Archiving
-75 years or as long as scientifically valuable
There are different types of computing with different requirements
Things working on:
-NOAA National Data Center
-NESDIS Enterprise Cloud WG
Metrics/Criteria for success
Pilots
-Amazon example
-prototyping a DAP server for the cloud environment
-elastic cloud computing - commercial and private cloud solutions and cloud storage, part II
Data storage strategic considerations
-expect 50pb in 5 years, 170pb by 2050
http://ocean-archive.data.nodc.noaa.gov//index.xml
Details on elastic computing strategic considerations
_______________________________________
Kai Liu: Cloud advisory tool
very quick slide presentation due to limited session time
Demo:
climateathome.com:8088/cloudadvisorytool/