Cloud Telecons 3

From Earth Science Information Partners (ESIP)

Dec 19, 2011 ESIP cloud computing cluster Telecon

  1. Participants: Carl Meyer, Ken Keiser, Rich Martin, Thomas Huang, Phil Yang, Doug Nerbet, Qunying Huang Aigun Heo, Aijun Chen, Brand Niemann, Liping Di, Long phan, Jason Simmons, Chris Webber, Peter Cornillon, Upendra Dadi, Thomson Nguy (Amazon)
  1. Co-Chair: Thomas Huang, Rick Martin, Phil Yang
  2. Supportor: Qunying Huang
  3. Call-in toll-free number (US/Canada): 1-877-668-4493
  4. Attendee access code: 23133897#


Agenda

  1. ESIP Winter Meeting Cloud Schedules (Thomas Huang)
  2. ESIP Cloud Computing Testbed (Phil Yang)
  3. Amazon cloud computing presentation starts around 2pm (Rick Martin)


Note:


ESIP Winter Meeting Cloud Schedules (Thomas Huang)

  1. Link for agenda: http://wiki.esipfed.org/index.php/Winter_2011_Agenda
  2. Talks from JPL : nebula, Amazon
  3. Talks from Nebula group: Nebula capabilities, architecture, open discussions for NASA cloud computing initiatives


ESIP Cloud Computing Testbed (Phil Yang)

  1. Requirements for cloud comptuing testbed have been identified:

http://wiki.esipfed.org/index.php/Cloud_Computing_Testbed_Requirements

  1. Discussed in the 2011 summer meeting: objectives, activities
  2. More topics will be discussed in the winter meeting
    1. The goal is to define the testbed, possible venders either to buy a testbed or contribute a testbed environment
    2. Computing resources from Azure (Carl and Phil)
    3. Nasa cloud resources for NASA users

Amazon cloud computing presentation (Rick Martin, Thomson Nguy )

  1. Prensenter: Thomson Nguy (Amazon)
  2. Topic: Overall of what Amazon doing, AWS clients doing, security
  3. Presentation Notes:
    1. Infrasturcture as a service
    2. Driving: on-demand easy to use, launch, turn off
    3. Customer choice: a variety of flexible choices for user, OS, programming tools, applications
    4. Flexible Pricing models: on-demand, elastic scale up and down, reserved instance or reserved fixed price, combination of price model , 50% saving price
    5. More than 100 governments and institution client
    6. Use cases: Mars exploration rovers, deep space network,carbon in the articic reservoir vuluneratiblity experiment, lunar mapper
    7. Mars Science Lab: fast motion field test- image processing in the cloud, massively parallel computations
    8. European space agency: ESA centre for earth observations; data collected by satellites in s3; 50.000 users at peak 30TB at a time; scale up storage
    9. Research and collaboration: Observable mdeical outcomes program(OMOP); Needed platform for cross-industry collaboration ; solution: OMOP use AWS for their research lab for providing a scientific computational platform for researchers; Benefits: avoided capital expense; flexibility to scale up and down based on computations demand and need; research tools can be shared with the extended community
    10. Top 5 Pharma: dynamic molecular modeling; large parallel problem; set up 30.000 cores –centos, 26.7 TB RAM, 2 PB disk space, HTTPs, ssh , 256-bit AES; supported software: condor(job management), chef (configuration management, Grill (cycle plugin for chef); speed and agility, $1.279 per hour, 3 hours batch job
    11. Security: AWS certifications, sarbanes-oxley(SOX), SAS70 type II audit, pci data security standard company (Credit card) etc;
    12. Security is a shared responsibility
    13. AWS VPC Architecture

QA

    1. Licence

Thomson: Buy licensed virtual machines charged by hours Doug: If you buy the agreement in advance, e.g, ESRI, you can use and install the vm

    1. How to select optimal options for choosing the VMs

Thomson: AWS provide the calculation model, but basically requires user to test and make decision

    1. Charge: any tools to monitor usage and costs

Thomson: a. create several accounts by IAM service, credit card/GSA , CloudWatch service to monitor the usage

    1. Computing units for cloud computing

Thomson: EC unit , ~ 1.0 -1.2 GHz

    1. Storage indexing for massive data

Thomson: S3, users bulid index; Amazon has DB services, simple DB, could help mange massive data;

    1. Customer back-up storage

Thomson: HSM, partners provide the tools for customizing the storage