Governance Issues Governance Dimensions of data access infrastructures

  • Slides: 21
Download presentation
Governance Issues Governance Dimensions of data access infrastructures Rob Atkinson Social Change Online

Governance Issues Governance Dimensions of data access infrastructures Rob Atkinson Social Change Online

Lessons Learned Design has to be informed by governance (policy implementation) but also by

Lessons Learned Design has to be informed by governance (policy implementation) but also by implementation Policy Cost-effectiveness of components established Governance Design Governance of components: • Catalogues • Common Models • Service Level Agreements Implementation Technology can be driven by policy AUKEGGS workshop, Edinburgh Sept 2005

Overview Business Realities Standards Lifecycles Semantic interoperability Within-domain consistency Cross-domain enablement AUKEGGS workshop, Edinburgh

Overview Business Realities Standards Lifecycles Semantic interoperability Within-domain consistency Cross-domain enablement AUKEGGS workshop, Edinburgh Sept 2005

Business Realities Scalable infrastructures must be adopted to be successful § One or two

Business Realities Scalable infrastructures must be adopted to be successful § One or two infrastructure paradigms per “node” § Current capabilities maintained § New ones trialled § Must be no significant barriers to adoption § Technical complexity of system vs simplicity of managing a node § Critical mass and community expectation ultimately wins AUKEGGS workshop, Edinburgh Sept 2005

Business Realities • Fear of change is overwhelming § Even when change is an

Business Realities • Fear of change is overwhelming § Even when change is an acknowledged necessity! • No solution is deployed, so all face scepticism • Needs to provide obvious value § But more importantly needs to be seen as the simplest route for individual players • Commitment of major players required § But within a governance framework that others are comfortable with AUKEGGS workshop, Edinburgh Sept 2005

Business Realities It’s the future that matters § Is it a business risk to

Business Realities It’s the future that matters § Is it a business risk to go a particular route? But the future is viewed through the present. . § Are safe, productive, easy steps being offered? § Is there a commitment and capacity to manage infrastructure? § Is it adaptable and scalable? AUKEGGS workshop, Edinburgh Sept 2005

Standards § Layer upon layer… § Encapsulate the “hard yards” into easy to use

Standards § Layer upon layer… § Encapsulate the “hard yards” into easy to use components § Easy to use only if the complete standards stack supported by quality implementations § Often multiple competing options § Usually due to poor separation of concerns § Typically forced to use a “poor” (ad-hoc) encoding because it’s the only semantic framework available… AUKEGGS workshop, Edinburgh Sept 2005

Standards Lifecycle § Existing practice § Identified need to harmonise and extend § Roadmap

Standards Lifecycle § Existing practice § Identified need to harmonise and extend § Roadmap § delegation to working group § Drafts § Revision § Testing § Adoption as baseline § Update strategy § Deprecation § Decommissioning AUKEGGS workshop, Edinburgh Sept 2005

Semantic Interoperability No über-model exists Communities of practice § Commitment and business drivers §

Semantic Interoperability No über-model exists Communities of practice § Commitment and business drivers § Re-usability and utility § Existence of a governance structure Semantics are declared not inferred § “This WCS serves MODIS data” § MODIS is not self-describing! Semantics implemented as a result of delegation § Delegate spatial semantics to OGC § IHO defines maritime navigation features § IHO delegates to ISO content standards § Etc AUKEGGS workshop, Edinburgh Sept 2005

Within-domain consistency Achieved by reuse of semantic elements § By declaration The good news:

Within-domain consistency Achieved by reuse of semantic elements § By declaration The good news: § Reuse is easy for lazy people § Productive people are “strategically lazy” § Reuse means power § Standard nuts and bolts means powered tools are economically feasible § Graphics libraries means games programmers don’t need to know much 3 D geometry Factor metadata around governance § Governance enables reuse! AUKEGGS workshop, Edinburgh Sept 2005

Practical considerations No rocket science Clearly delineate roles with respect to necessarily reusable artefacts

Practical considerations No rocket science Clearly delineate roles with respect to necessarily reusable artefacts § The entity creating a data server might be the person who defines the data specification § OGC specs tend to assume it is (IMHO a simplification that causes a complex problem) § But in general, interoperability demands that the definition is agreed in a broader community AUKEGGS workshop, Edinburgh Sept 2005

The “USER” will “DISCOVER” “DATA” and “ACCESS” it. Means a lot of work somewhere…

The “USER” will “DISCOVER” “DATA” and “ACCESS” it. Means a lot of work somewhere… We can model (and implement!) what is required AUKEGGS workshop, Edinburgh Sept 2005

AUKEGGS workshop, Edinburgh Sept 2005

AUKEGGS workshop, Edinburgh Sept 2005

What it means Infrastructure must: § Define (or adopt) component standards (schemas for metadata

What it means Infrastructure must: § Define (or adopt) component standards (schemas for metadata artefacts) § Deploy capable components § Define governance arrangements – how are these roles assigned § Be a viable business proposition to participate § As one of a very few active participations! § Broker interoperability with other infrastructures § Have a strategy to adapt and adopt new expectations AUKEGGS workshop, Edinburgh Sept 2005

Case study: biological observations Typical query: § What threatened species have been recorded in

Case study: biological observations Typical query: § What threatened species have been recorded in area X? a “join” is required between separately managed sources of information: § Taxonomy – terms, synonyms and hierarchy § Grouping – e. g. status according to Schedule A § Records § Spatial constraint AUKEGGS workshop, Edinburgh Sept 2005

Join-in-situ at Data Access Service Grouping type Data Access Service Term(s) Term List (preferred)

Join-in-situ at Data Access Service Grouping type Data Access Service Term(s) Term List (preferred) Groupings Synonyms Used Record Set (with synonyms, Groupings, Preferred terms) Records AUKEGGS workshop, Edinburgh Sept 2005

Query. . Grouping type Term(s) Taxonomy Preferred Term(s) (for grouping) Grouping Preferred Term(s) Taxonomy

Query. . Grouping type Term(s) Taxonomy Preferred Term(s) (for grouping) Grouping Preferred Term(s) Taxonomy Terms and synonyms Register Local synonyms Data Access Service Record Set (with inconsistent local synonyms only) AUKEGGS workshop, Edinburgh Sept 2005

Response… Record Set (with synonyms) Extract terms Term(s) Taxonomy (Grouping) Taxonomy Service Term List

Response… Record Set (with synonyms) Extract terms Term(s) Taxonomy (Grouping) Taxonomy Service Term List (with preferred terms identified) Sort results Groupings per term If results to be presented according to grouping – e. g. Threatened, Vulnerable etc Presented Results AUKEGGS workshop, Edinburgh Sept 2005

Over Hurdle 1…. . Automate update of terms, from point oftruth services: update Grouping

Over Hurdle 1…. . Automate update of terms, from point oftruth services: update Grouping harvest Taxonomy Service Client Term List (preferred) Groupings Synonyms Used Taxonomy Records AUKEGGS workshop, Edinburgh Sept 2005

Wrappers around “immutable” data sources: Grouping type Term(s) Wrapper High-bandwidth link Record Set (with

Wrappers around “immutable” data sources: Grouping type Term(s) Wrapper High-bandwidth link Record Set (with synonyms, Groupings, Preferred terms) Term List (preferred) Groupings Taxonomy Service Client Synonyms Grouping Taxonomy Data Access Service Synonyms used A vocab from data Service – must match Synonyms in taxonomy Records AUKEGGS workshop, Edinburgh Sept 2005

Solution success criteria üMeets requirements ücan be extended to encompass more data services ücan

Solution success criteria üMeets requirements ücan be extended to encompass more data services ücan be extended by new ways of looking at the data üdata is as simply re-usable as possible ühas clear and achievable governance requirements üpromises reasonable performance ücan be implemented in a partial solution initially üprovides mechanisms for improving data consistency over time AUKEGGS workshop, Edinburgh Sept 2005