HPCLocMtx

From HPC Wiki
Revision as of 17:31, 25 February 2019 by Hpcwiki dept.admin (Talk | contribs)

Jump to: navigation, search

HPC Location Matrix

This matrix lists the pros and cons of types of HPC hardware ownership and hardware locations.


HPC Location Matrix
Item Category Subcategory A Subcategory B Pros Cons Notes
1 Hardware owned
by institution
2 On-campus
3 Strongly preferred by NJIT researchers using HPC Hardware will not stay current without funded timely re-fresh
4 Provides maximum control for researchers and administrators Expansion is limited by physical space, and power and HVAC requirements
5 HPCBaseline provides guaranteed capability for researchers without and with their own funding Expansion of resources takes months, compared to typically less than 1 hour for cloud
6 Researchers who own hardware have continued use of that hardware when their funding runs out
7 For HPC use patterns, less expensive than commercial cloud
8 Easily integrated into the NJIT AFS software environment
9 Can be expanded into the cloud as needed and funded
10 Colorado School of Mines (peer institution) stance
11 Campus Champions On/off-premise HPC See: 1)Last bullet in "Section 1 responses"
2) Next to last bulle tn "Section 2 responses"
12 Off-campus
13 Co-location
14 Differences from "On-campus": Item 4, Diminished control for administrators; Item 8, Integration may be problematical Differences from "On-campus" None
15 Commercial
16 Differences from "On-campus": Item 4, Significantly diminished control for administrators; Item 8, Integration likely to be problematical Differences from "On-campus" Items 3, 4, 5 do not apply Costs are unlnown
17 Hardware not owned
by institution
18 Free cloud [1]
19 "Evergreen" - new processors readily available Successful proposal needed
20 Flexible configuration Access ends when proposal period ends
21 Expandable, within propsal limits Loss of common software environment compared to owned
22 Commercial cloud [2]
23 "Evergreen" - new processors readily available Generally more expensive for HPC than owned hardware
24 Flexible configuration Loss of common software environment compared to owned
25 Rapidly expandable Access ends when funds run out

Legend
[1] E.g., XSEDE
[2] E.g., AWS, Azure, GCP, Penguin