Den

1
Open source for statistics A stairway to
heaven? Joint ECE/Eurostat/OECD Meeting on the
Management of Statistical Information Systems
(MSIS) Geneva, 17-19 May 2004 Rune
Gløersen Director of IT and Data
Collection Statistics Norway
2
Topics covered

• A general discussion on strategic aspects
  concerning Free/Libre/Open Source Software
  (FLOSS)
• A short presentation of the purpose, current
  state and possible future for the Statistics Open
  Standards Working Group
• Conclusions (if any)

3
Aspects of open source applications

• Best practice. The Linux and Perl revolution has
  prepared the way for share and reuse best
  practice related to version management, testing,
  security and reliability assessment. When
  collaboration projects (running software pooling)
  are conducted by such practice, optimized
  investments may be derived
• Interoperability and de facto standards. The
  successful dissemination of open source
  platforms, enforce main vendors to provide for
  this platform to be in business. E.g. SAP,
  Intentia and Oracle are well off on their
  portfolio adjustments to Linux

4
Aspects of open source applications (2)

• Quality and flexibility. Commercial trusts and
  consortiums seem to overtake the leadership of
  open source development models. The combination
  of an emerging new business model and deep
  marketing understanding, provides a driving force
  for high quality software and open standards
  (e.g. Sun, IBM and Novell)
• Security. As proprietary software is hidden,
  security inspection is impossible. The open
  source possibility of code inspection, and
  security gap encountering is about to fortify in
  military software applications.
• Economics. Organizations where labour and
  support costs are low, as the public sector, tend
  to have eager plans for open source software.
  This is due to the fact that license fees will
  turn to be a more significant cost factor in the
  budgets

5
Aspects of open source applications (3)

• IT management skills. When the open-source
  discussion is not formally addressed, its likely
  that the developers introduce open source
  products in an ad-hoc manner. Then minimum
  skills required should be
• The cost balance balancing the cost and skills
  required for proprietary upgrades vs. the cost of
  running a heterogeneous IT-portfolio (including a
  mix of free and commercial products)
• The security and reliability requirements of the
  organization awareness of the potential
  prioritized areas of open source appliances
  related to such requirements

6
Aspects of open source applications (4)

• Developers skills. The success of open source
  projects, very much depends on utilizing best
  practice techniques
• The governmental intentions. In broad outline,
  the European countries seem to have the same
  intentions of open software applications
  interoperability, security and cost reduction.
  The statistical sector, as a whole, is perhaps
  not in the position to draw upon specific
  advantages of FLOSS due to the lack of a common
  business model widely used

7
The Statistics Open Standards (SOS) Group

• Established 1998 as a Nordic IT Management Group
• Members NSIs (IT function) in Denmark, Sweden,
  Finland, Iceland and Norway.
• Broadened 1999 by
• scope Statistics Open Source
• members CBS Netherlands and the Swiss
  statistical office included
• An initiative to establish a forum of common
  interest in SW development close to the NSIs need
  and strategy

8
Some main principles

• Exchange of information (frankly speaking!)
• Re-use components
• Keep single-owner responsibility
• No formality
• No common funding
• Two levels
• technical level
• Steering Committee, have met approx. twice a year.

9
Common projects

• Swe/de/nor development and use of common
  dissemination database (Statistical Databank)
• Neuchâtel Classification Model
• Dutch Cristal model
• Gesmes data manager
• Investigation and purchase of SuperStar II
• XML
• XML standards/interfaces
• data collection
• Coming up web-services

10
Some experiences

• Common projects close to NSIs needs make the best
  examples of cooperation
• Difficult to bridge differences in technology,
  architecture and knowledge.
• More or less hidden barriers in
  projects/products, which at a glance seem to be
  open and easy to implement
• Clustering used to enhance bargaining
  power.SuperStar as an example
• General appreciation of the informal exchange of
  information

11
Some check-pointswhen considering FLOSS

• License conditions of your proprietary software
  (extent of source code access, reproduction,
  interoperability, amongst others)
• Matters of exceptional usage, when proprietary
  software license conditions will not be fulfilled
• Preferred application areas within the
  organisations infrastructure (i.e. security and
  reliability applications, internet applications)
• Organisational skills requirements
• Adaptation to governmental overall strategy, if
  any

12
Some conclusionswhen considering joint venture

• Well defined interfaces in terms of adopted
  international standards or commonly agreed
  standards are key to success in any type of
  cooperation of this kind
• While a strategy based on discussing needs, and
  then jumping to software development has shown to
  be a failure, mostly any cooperation based upon
  defined interfaces has shown to be a success
• Pay attention to internal barriers for adoption
• Technological
• Architectural
• Organisational
• Human knowledge