150 likes | 269 Views
Data Management METRICS for NNDC and CLASS. David Hermreck. Metrics - Context. Revisit appropriate operational performance metrics in an environment with an operational CLASS. CLASS and NNDC metrics are currently overlapping. Metrics should focus on core functions.
E N D
Data Management METRICSfor NNDC and CLASS David Hermreck DISCUSSION DRAFT ONLY
Metrics - Context • Revisit appropriate operational performance metrics in an environment with an operational CLASS. • CLASS and NNDC metrics are currently overlapping. • Metrics should focus on core functions. • Need both Development and Operations Metrics • Will use a “bank” analogy to better understand CLASS and NNDC Operations roles DISCUSSION DRAFT ONLY
CLASS Development Metrics • CLASS development requires Capability Maturity Model Integration (CMMI) level 3 – this model provides many potential metrics • Potential metrics could include: • Number of Change Requests implemented on time and budget • Major software releases delivered to operations on time and budget • Other CMMI metrics DISCUSSION DRAFT ONLY
Operations Metrics - The Bank Analogy CLASS Safe and Secure Preservation is… is NOT… just storage DISCUSSION DRAFT ONLY
The Bank Analogy CLASS Safe and Secure Preservation is… is NOT… Stewardship DISCUSSION DRAFT ONLY
The Bank Analogy CLASS Wholesale Access is NOT… is… Retail Access DISCUSSION DRAFT ONLY
The Bank Analogy CLASS is… NNDC is… Primary “Interbank” Transactions • “Owner” Deposits • “Owner” Withdrawal Commercial Access • “Read Only” • Expert Users Retail Access: • Checks • Debit Cards • Branch Banking • ATM access • Public usability DISCUSSION DRAFT ONLY
Ocean Perspective Geophysical Perspective Climate Perspective NODC NGDC NCDC • CLASS Services – • Ingest • Archive Storage • “One NOAA” Access • Coordinated OAIS with NNDC CLASS Service Domain Expert Layer Stewardship Layer IT Services Layer Preservation Layer Raw Ingest Raw Access • XML? • Per submission agreements DISCUSSION DRAFT ONLY
Service Metrics Ocean Perspective Geophysical Perspective Climate Perspective • Customers Served • Tailored portals • Stewardship actions • Metadata enhancement Domain Portals & Expert User Support NODC NGDC NCDC Domain Expert Layer Stewardship Layer IT Services Layer Preservation Layer • CLASS Metrics – • Data Stored (PB, # sets) • Data Accessed (# inquiry, volume) • Data Ingested • Latency • Preservation Activity Raw Ingest Raw Access • XML? • computer to computer • Per submission agreements Note that NNDC stewarded (“owned”) data can be delivered to end-users without passing through the NNDC owner’s site. DISCUSSION DRAFT ONLY
NNDC Metrics: Questions re. CLASS • Does CLASS support: • subsetting capabilities? (Does this require an inappropriate understanding of content?) • data mining? • Regeneration (e.g., producing an intermediate data form in response to a query?) • Can CLASS change (obsolete) the external “look and feel” of data access (e.g., no command line access)? • Can CLASS obsolete “old” access methods (e.g., dial-up modems, 8” diskettes, 2” tape, etc.?) • How do these impact CLASS metrics? DISCUSSION DRAFT ONLY
NNDC Metrics: Questions re. CLASS • Is CLASS independently responsible for reversible transformations (perhaps as part of media migration)? Is this an “operations” question? • Can CLASS independently do irreversible transformations, if required for data preservation? Can CLASS move obsolete editions to lower care levels? • How does CLASS measure closely coupled data transfers (e.g., for reprocessing)? • How do these impact NNDC metrics? DISCUSSION DRAFT ONLY
Metrics • CLASS metrics would move toward: • infrastructure/wholesale • storage and preservation • NNDC metrics would move toward: • Value-added/retail • Stewardship focused • CLASS metrics and NNDC metrics should (eventually) be distinguishable. • However, CLASS should report nonintermediated access statistics for NNDC owned/stewarded datasets. • Metrics need further development. DISCUSSION DRAFT ONLY
CLASS Metrics ?? • Volume in and out by time, by NNDC • Volume stored • Collection Inventory • Inventory changes • Preservation activity • Data flows & latency • Storage and bandwidth reserves DISCUSSION DRAFT ONLY
NNDC Metrics ?? • Number & quality of Value-added interfaces (usage?) • Datasets reprocessed or enhanced • # datasets at highest level of maturity • Volume served/total of data at highest level of maturity • Customer liaison contacts • Metadata enhancements DISCUSSION DRAFT ONLY
Conclusion • Eventually, NNDC and CLASS metrics should be distinct. • More work is needed to identify the “right” metrics to measure effectiveness. • Coordinate with other data centers in NASA and USGS on metrics • Good metrics are HARD! – multiple measures are required DISCUSSION DRAFT ONLY