March 14, 1999
White Paper - Consolidated Allocation/Utilitzation Database – by John Baird
– only for allocated projects (which includes Challenge Projects)
Service/Agency – and associated Service or Principal and designate(s)
User organization – and associated S/AAA(s), an S/AAA may be associated with more than one user organization
Project – and associated Principal Investigator, CTA, Challenge status, and other project information
Subproject (one blank instance if no subprojects) – and list of co-investigator/user(s) for subproject
HPCMO resource (what architecture at what specific SRC or may be across set of SRCs)
Year-end hours allocated in prior FY (if project and subproject existed in prior year)
Initial hours allocated in current FY
Current hours allocated in current FY (may be derived from deltas and initial allocation)
Historical record of each adjustment in hours allocated during current FY (delta, when made,
who made change, and where delta came from/went to, comments if needed)
HPCMO – at start of FY, to roll current allocation into prior FY, clear history, and set new allocation for old and new projects
S/AAA – to make re-allocations for any Project under and across any User organization(s) they are associated with. Jeanie Osburn, POC
Principal or designate(s) – to make retrievals for any Projects under their Service or Agency
S/AAA – to make retrievals for any Projects under any User organization(s) they are associated with. Jeanie Osburn, POC
PI – to make retrievals for any Projects associated with their name
User – to make retrievals for any Subprojects (or Project if no Subprojects) associated with their name
HPCMO staff – to make retrievals across entire allocation data base
HPCMO staff – to make corrections, with comments
– for all allocated projects (which includes Challenge Projects) as well as "special" HPC accounts
– for "special" HPC accounts, User organization and S/AAA will have some special value [blank?]
– for "special" HPC accounts like maintenance and down time, there may be no associated users
Service/Agency – and associated Service or Principal and designate(s)
User organization – and associated S/AAA(s), an S/AAA may be associated with more than one user organization
Project – and associated Principal Investigator, CTA, and other project information
Subproject
Co-investigator/user for subproject
HPCMO resource – what architecture at what specific SRC
Queue type – allocated, background, etc.
Total hours utilized in prior FY (if project and subproject existed in prior year,
and user was associated with subproject or project)
Current total hours utilized in current FY to date (may be derived by summing deltas)
Historical record of each utilization delta for current FY – delta, when made, and
comments if needed (on negative deltas, for example; also who made comment)
important: Should only record deltas day by day, not down to individual job.
HPCMO – at start of FY, to roll current utilization into prior FY, clear history and current utilization for all projects
Principal or designate(s) – to make retrievals for any Projects under their Service or Agency
S/AAA – to make retrievals for any Projects under any User organization(s) their name appears under
PI – to make retrievals for any Projects associated with their name
User – to make retrievals for any Subprojects (or Project if no Subprojects) associated with their name
HPCMO staff – to make retrievals across entire utilization data base
HPCMO staff – to make corrections, with comments
– data updated monthly, and only updated once each month
– for all allocated projects (which includes Challenge Projects) as well as "special" HPC accounts
– for "special" HPC accounts like maintenance and down time, there may be no associated user or S/AAA
– data is kept for the entire prior FY and current FY to date, by month
- total for prior FY and for current FY to date are kept (may be derived)
Number of active users with >1 CPU hour utilization in the given month
Number of cumulative users with >1 CPU hour utilization in any month, FY to date
Data for unnormalized expansion factors
Data for normalized expansion factors by queue
Data for normalized expansion factors by Challenge Project
COTS software utilization
1. the number of accesses to each program or library (program executions and library link references)
2. the CPU-hours used on each COTS program. (Programs only, not for COTS libraries)
3. the number of load references for dynamic libraries
HPCMO – at start of FY, to roll current data into prior FY, clear history and current utilization for all projects
HPCMO staff – to make retrievals across entire other utilization data base
HPCMO staff – to make corrections, with comments
Interface with external data base at HPCMO to provide information to be used in generating/updating Account Applications (which are stored in the external data base, not as part of allocation/utilization data base. Larry Davis, POC
Data and data currency requirements unknown separate/consolidated with Allocation/Reutilization data base unknown. Consolidation suggested by Help Desk Statistics/Problem Tracking collaboration team 6/98
Serve as a common shared data base of trouble tickets to support cross-Center customer service support of users (e.g., ticket opened at one Center, acted on by another). Possibly needed as part of metacomputing implementation across Centers. Frank Lovato, POC
Note: Cherri Pancake, Oregon State University and NAVO PET, has a pilot for Web-based resource accounting, and needs to be included in requirements definition.