Collection Architecture 2009 MSIS Meeting – Oslo, Norway Karen Doherty May 18, 2009
Statistics Canada Statistique Canada 2 Agenda Collection at Statistics Canada Business Architecture Initiative IT Architecture of Collection systems Lessons Learned
Statistics Canada Statistique Canada 3 Collection at StatCan HQ (0ttawa) plus eight regional collection hubs (Halifax, Montreal, Sherbrooke, Toronto, Sturgeon Falls, Winnipeg, Edmonton, Vancouver) Collection methods Face-to-face interviewing (CAPI) Telephone interviewing (CATI & Random Digit Dialing) Paper questionnaires (Canada Post) Electronic data reporting (Internet) Call Centre (mainly for Census)
Statistics Canada Statistique Canada 4 Business Drivers Cost: Budget cuts => need to reduce costs Telecom: Canadian moving to cell phones => hard to reach Electronic Reporting: Canadians (especially younger ones) want Internet services Geography: Large tracts of land (remote areas, far north, etc.) will continue to be a challenge Admin Data/Registries: Use is increasing Multi-mode Collection: Flipping between modes is more and more common
Statistics Canada Statistique Canada 5 Business Architecture Initiative StatCan embarked on a business architecture project in partnership with a private sector firm Input sought from a wide range of stakeholders Looked at business drivers, requirements, services, responsibilities, organization Restricted to collection activities Excluded Census but Census requirements considered during systems specification stage A lot of push back and scepticism at the beginning even from very senior management
Statistics Canada Statistique Canada 6 BA Recommendations Provide respondents with a menu of response options Realign the organization to match the business delivery model Separate operations from development Adopt common services, processes and tools Invest in new technology (corporate IT services, wireless, thin client, etc.) Develop new collection applications
Statistics Canada Statistique Canada 7 Survey Master Control System Functionality includes: Management of case load regardless of collection mode Movement of cases between modes Transfer of case data to and from legacy stand-alone collection applications (until they are replaced) A data repository for all cases Reporting tool for all collection activity
Statistics Canada Statistique Canada 8 Survey Master Control System
Statistics Canada Statistique Canada 9 Survey Master Control System Designed by the in-house architecture team using a Service Oriented Architecture model Technology StatCan’s MS.net Development Framework MS Biztalk integration platform MS SQLServer database Development outsourced to a private sector firm Production in 2009
Statistics Canada Statistique Canada 10 e-Questionnaire & e-File Transfer e-Questionnaire: Internet-based questionnaires (similar to what was used for 2006 Census) Production in 2009 e-File Transfer: Managed bi-directional movement of files between StatCan and respondents, clients, users Production in 2008 COTS products acquired via tender Access control & full end-to-end encryption
Statistics Canada Statistique Canada 11 E-Questionnaire & e-File Transfer
Statistics Canada Statistique Canada 12 Other Tools Generalized Tracking System Tracking of paper questionnaires, supplies, etc. In-house developed MS.net application Production 2009 Case Management Requirements definition in 2009 Field Management System Developed for Census In use for 2009 Census Test Will be adopted by survey Field Staff
Statistics Canada Statistique Canada 13 Adherence to StatCan EA All applications run on the standard suite of infrastructure services All development done using standard development frameworks Requirements and documentation managed via the IBM Rational Suite (Rational Unified Process) and Team Foundation Server Heavy use of IBM automated testing tools
Statistics Canada Statistique Canada 14 Lessons Learned Take the time to review the business architecture (question everything, challenge the status quo) Need strong governance (senior management support) Need effective change management process throughout COTS products can be expensive and take a long time to procure but implementation is quick if you don’t customize much Use of common tools speeds up development but must make sure resources are properly trained