Server and Short to Mid Term Storage
Funding
Research Computing Funding Issues
Stages of Development
• Starting• Building• Sustaining
Where are you computing?
• Research Group• Department• Institution• Region• National Laboratory/Teragrid• Cloud
What are you funding?
• Computational systems• Storage– Scratch– Fast working data– Medium term
• Software• Visualization
Funding Sources• Individual– Grants– Startup funds
• Departmental– Grants– Departmental funds
• Central– Grants
• Institutional• Multi-institutional
– General funds
Central Funding Models
• Centrally operated• Condo• Fee for service
Life Cycle Management
• Decommission on a pre-determined schedule• Decommission when a replacement is
purchased• Decommission based on performance criteria
(ie – FLOPS/Watt compared with current technology)
• Run until the wheels fall off• Other
Total Cost of Ownership
• Do purchasing decisions take total cost of ownership into consideration, or are systems purchased based purely on researchers needs/desires?
• At what point does the central IT staff get involved in the acquisition process?
SURVEY RESULTS
4. Computational hardware resources (clusters, SMP, viz center):
# Answer Response %
1 NO 0 0%
2 Yes, centrally run 16 80%
3 Yes, departmentally run 17 85%
4 Yes, run otherwise 3 15%
Yes, run otherwise (specify)
cloud (Amazon, Azure, Google/IBM)
Computational Institute provides terragrid
grants
5. Are your centrally run hardware resources:
# Answer Response %
1 Funded by charge back 5 31%
2 Funded centrally 13 81%
3 Funded otherwise 11 69%
Funded otherwise (specify)
grant funded + dept'l contribution (e.g. faculty start-up)
Condo model - faculty contributing research funds
"condo" model: cluster infrastructure funded by sponsors, blades purchased by researcher
"Buy-in" program to permit faculty to use grant income to buy nodes in central facility
combination of both
Grant Funded
grant funded
Grants
research grants can purchase machines and locate them in the central cluster
grants
Grants
6. Are your departmentally run hardware resources:
# Answer Response %
1 Funded by charge back 6 35%
2 Funded departmentally 16 94%
3 Funded otherwise 9 53%
Funded otherwise (specify)
grant
power is funded centrally
Grant income
grant funded
I am not sure how funding is provided for by the divisional/departmental provided hardware resources but for those resources provided by the researchers they are funded from their grants, contracts, and/or awards.
Grants
research grants
grants
infrequent creative supplements by dept
7. Computational research software:
# Answer Response %
1 NO 0 0%
2 Yes, centrally run 17 85%
3 Yes, departmentally run 17 85%
4 Yes, run otherwise 5 25%
Yes, run otherwise (specify)
Individual PIs
My understanding is that the researchers run their own research software since it really is specific to their field of expertise. There are cases where software is installed and supported by the departmental staff, again citing the Computational Institute as one such offering.
grants & centrally managed site licenses
research groups
collaborative licensing on general purpose SW
8. Are your centrally run software resources:
# Answer Response %
1 Funded by charge back 5 31%
2 Funded centrally 16 100%
3 Funded otherwise 3 19%
Funded otherwise (specify)
combination of eScience Institute, and "condo" sponsors
grants
Grants
9. Are your departmentally run software resources:
# Answer Response %
1 Funded by charge back 5 29%
2 Funded departmentally 13 76%
3 Funded otherwise 5 29%
Funded otherwise (specify)
grant-funded
Grant income
My assumption is that funding comes directly from the researcher through their grants, contracts, and/or awards.
grants
Not sure
Summary of Recommendations
Developing a Coherent Cyberinfrastructure from Local Campus to National Facilities:
Challenges and Strategies
EDUCAUSE Campus Cyberinfrastructure Working GroupCoalition for Academic Scientific Computation
Workshop: July 2008Report: February 2009
§2.1: Harnessing Campus and National Resources
• SR 2.1.1: Campuses, in partnership with national resource providers and governmental agencies, should support, promote, and develop a coherent, comprehensive set of computing and data facilities.– TR 2.1.1a: Integrate national resources with the campus layer
in a way that ensures transparency, scalability, and ease of use.– TR 2.1.1b: Develop funding models that enable and demand
integration of resources from lab to campus to national center.– TR 2.1.1c: Develop and deploy processes and policies that
ensure flexibility for PIs to choose local or national resources.
§2.1: Harnessing, cont’d• SR 2.1.1, continued– TR 2.1.1d: [Federal and state] governmental funding
agencies should implement [award] terms that encourage sharing and effective use of resources at all layers, while eliminating disincentives for researchers to use campus or other shared resources.
– TR 2.1.1e: Campuses should encourage resource sharing where local governance and policy allow it, thus helping improve scholarship.
– TR 2.1.1f: Campus IT organizations should take an active role in exploring new technologies by serving as a conduit via the CIO (or equivalent) to promote and develop new capabilities and access to resources that are external to the campus.