Cloud Computing Testbed Requirements

Intro

A progress report and discussion about what is needed to support a cloud computing testbed for producing lessons learned and best practice for ESIP members to assess, test, and adopt cloud platforms for their research, development, and education:

• discuss the matrix of requirements for cloud computing by ESIP members,
• identify the possibility of cloud resources (Amazon, Microsoft, Nebula, Private Cloud),
• identify potential applications to be running on potential cloud resources,
• a execution plan for the testbed with a target of presenting in the summer meeting.

Earth Science Information Requirements and Computational Characteristics

• Data Intensive (Disk I/O, Network I/O)
• Computing Intensive (Numerical Models)
• Concurrent Intensive (Popular Services)
• Cloud Virtual Hosting
• Intensity Spikes
• Scalability/Reliability (guarantee % of up time)

Vendor Capabilities

• Scalability/Reliability (guarantee % of up time)

• Maximum or Available Capacity
  • Operation
  • Spike
  • CPU Core
  • RAM
  • Storage
  • Network (In-bound, Out-bound)

• Cost Model (how vendor charge the following aspects) and Cost
  • Operation
  • Spike
  • CPU Core
  • RAM
  • Storage
  • Network (In-bound, Out-bound)
  • Other Facility (e.g., auto scaling, queue, logs, analyses, etc.)

• Available User Interfaces
  • Web Interface
  • SSH Linux type of interface
  • Or community open APIs

• Supported Services
  • SaaS, PaaS, IaaS, etc.
  • Supported OS
  • Apache Hadoop
  • Persistent object store
  • Intercloud

• Usability and Manageability

• Security and Auditability (FISMA supports)

• References supporting other science-based orgs
  • Successful Domain Projects
  • Deployed science-domain projects using their platform