Transcript SISS_WxS_Workshop_-_GSV_Experience

Deploying OGC Web Services
GeoScience Victoria’s Experience
Alistair Ritchie, Senior Information Geologist
GeoScience Victoria
Outline
• Background
• GeoScience Victoria services
• Application schema and technology used
•
•
•
•
•
deegree web feature service overview
Geoserver ‘web-service’ overview
Comparison of deegree with Geoserver app-schema
The future
Observations
Background
GeoScience Victoria (GSV)
• Victorian state geological survey
• Custodian of state’s geology and mineral exploration data
Why use web services and community schema?
• Maintain a number of systems which change over time
• Need a stable standard interface to data
• Need a context in which to improve data quality
History of community involvement
• Joined CGI interoperability working group in 2004
• Joined EarthResourceML working group in 2007
• Close collaboration with AuScope/ASRDC since 2008
GSV Services – app. schema
GeoScience Markup Langauge (GeoSciML)
•
•
•
•
•
Deliver detailed geological map data
Deliver results of drilling activities and resulting sub-surface sampling
Participated in the development of the language and web services
Geology services used by OneGeology geological map of the world
Boreholes services support the National Virtual Core Library
Earth Resource Markup Langauge (EarthResourceML)
• Deliver descriptions of mineral deposits and associated mining activities
• Participated in the development of the language and web services
• Collaboration with AuScope to deploy Australian network of services
GSV Services - Technology
deegree v2.*
• GeoSciML v2.0 – testbed and production services
• EarthResourceML v1.1 – production
• GeoSciML v2.0 (boreholes) – testbed and production services
deegree v3.0
• Very little experience – German demo serving GSV data
• Candidate for GeoSciML v3.0 testbed
Geoserver app-schema
• EarthResourceML v1.1 – test services (soon to replace deegree)
GeoServer web-service
• GeoSciML v2.0 – in development to replace deegree
deegree overview
•
•
•
•
•
Java OCG Web Server Application
Developed at the University of Bonn, Germany
Use stable release - version 2
Conforms to OGC WFS 1.1 specification
Supports GML 3.1 app. schema ... almost ...
• Only one namespace
• No support of XML attributes
• All complex types must be «features» (no «types» or «data types»)
• Has an XSLT layer to overcome this
• Transform ‘private’ GML to GeoSciML (request/response)
• Implement data-type polymorphism
deegree overview
<<xml>>
GeoSciML
Request
Client
<<gml>>
GeoSciML
Response
Virtual
outputFormat
<<xslt>>
inFilter
<<xslt>>
outFilter
Deegree WFS
<<xsd>>
Private
Schema
Datastore
Geoserver web-service
•
•
•
•
•
Web service datastore for app-schema
Adds the OGC WFS interface to a SOAP web service
Developed by an AuScope funded embedded software engineer
Binds to a ‘generic’ SOAP web service
Xpath mapping of nodes in SOAP service response to nodes in
the target schema
Geoserver web-service
<<xml>>
GeoSciML
Request
Client
<<gml>>
GeoSciML
Response
GeoServer
web-service
request
mapping
response
mapping
GSV WS
(SOAP)
<<xsd>>
Private
Schema
Datastore
(Hibernate)
deegree–Geoserver comparison
deegree 2.n
Geoserver app-schema
polymorphism
+
+
flexibility
+
-
documentation/support
-
+
query capability
-
+
configuration
-
+
performance
+
-
deegree–Geoserver comparison
gsml:MappedFeature response times
Response time (s)
80
70
60
50
deegree
40
Geoserver
30
20
10
0
0
50
No. Features
100
deegree–Geoserver comparison
•
•
•
•
XSLT makes deegree 2 a flexible testbed tool
Deegree 3 - an unknown quantity but promising reviews
Geoserver is a complete WFS 1.1/GML3.1.n package
Geoserver’s performance is a major drawback
Future work
GSV’s Geoserver wish list
•
•
•
•
•
•
•
•
User interface
WFS 2.0 support
Deploy multiple service end-points from one Geoserver instance
Improved Web Map Service integration
Data mapping and transformation tools – no database views?
WFS-T for app-schema datastores
3D support (response and request)
Performance must improve
Observations
Know your database
• Must be well designed and configured
• Don’t neglect spatial and aspatial indexes
Know your servers
• WFSs are like any other web app – deploy accordingly
• A good system administrator is vital
Honour the specifications
• Community schema and profiles
• Web service specifications
• Either 100% compliant or not at all
Observations
The hard work is mapping data models
•
Majority of the deployment time is taken up here
1. Mapping corporate data models to the community schema
2. Mapping local vocabularies to community vocabularies
3. Physically realising the mapping in the data source
The hardest work is social
•
•
Technology is no longer an impediment
Corporate IT service providers can be
Observations
Corporate IT support is crucial
• At the very least they provide the holes in firewalls
• Provide stable domains for URIs
•
•
•
•
•
Machine to machine infrastructure – machines are stupid
Links to services must not break
Links between services must not break
Government department domains are bad for service URIs
Significant bureaucratic effort required to acquire alternative
Start working with corporate right from the start
Questions?