Science Data Access Architectures

1
Science Data Access Architectures Mike
Martin 11/30/06
2

• Background
• Proposed to Ed to survey information system
  capabilities across NASA disciplines. Where are
  they now, where are they going?
• Especially interested in capabilities and tools
  that could answer requirements in MRO User
  Scenario Document by Ray Arvidson and Jim Murphy,
  something of a PDS vision statement.
• First step, look at middleware architectures.

3

• For several selected middleware architectures
• Describe how each scheme works.
• Compare the different approaches to common
  problems.
• Identify any exemplary or noteworthy features
  of the various schemes.
• It quickly became apparent that there was a lot
  of excitement about Service Oriented Architecture
  (SOA) and Web Services.
• Commoditize information and computer resources
  for autonomous consumption.
• SOAP messages, WSDL service contracts and UDDI
  service registries.

4

• Selected Systems include
• Open GIS Consortiums Open Web Services (OWS),
  servers for digital maps, images and GIS data.
• International Virtual Observatory Alliance
  (IVOA) registry and data access services for
  astrophysics data.
• Planetary Data Systems implementation of
  Object Oriented Data Technology (PDS-OODT) for
  query, profile and product servers.
• Earth Observing System Clearinghouse (ECHO) for
  a central catalog identifying earth science data
  and services.

5
Open Web Services (OWS)
6
International Virtual Observatory Alliance (IVOA)
7
PDS Object Oriented Data Technology (OODT)
8
EOS Clearinghouse (ECHO)
9
Feature Comparison
10
Service Summary
11

• Highlights
• OWS layered discovery model (GetCapabilites(),
  DescribeRecords(), GetRecords() and harvesting
  capability
• IVOA cone search, VOTable, summer studies.
• Web Services orientation in ECHO and IVOA.
• Lowlights
• No standard XML query language
• Mixing service registry and catalog functions.

12

• Conclusions
• OWS - Elegant architecture. Now someone needs
  to make it work, especially the catalog servers.
• IVOA - Great architectural vision.
  Interesting data oriented service approach.
  Seems to have widespread community support.
• PDS-OODT - Great job implementing product
  servers, now do profile servers at every node.
  IPDA efforts are adding support for VOTable and
  GetCapabilites features.
• ECHO - Wide range of services exposed via
  API. Still seems immature, despite being on
  Version 8. Will central spatial database be a
  success?

13

• Next steps
• Get feedback from you folks and others at
  tahoe_mike_at_sbcglobal.net
• Should other schemes be included (the Grid,
  Opendap)?
• Other analysis/observations based on this
  information?
• Is there hope for long term convergence (earth,
  space, planetary and astrophysics all using same
  service architectures)?