32
Guide to sharing statistical services in the European Statistical Service Sharing Common Functionalities in the ESS (SCFE) Joni Karanka, ONS

Guide to sharing statistical services in the European Statistical … · 1. Before we start •The identification, definition and adoption of statistical services is not an IT-only

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

  • Guide to sharing statistical services in the European Statistical Service

    Sharing Common Functionalities in the ESS (SCFE)

    Joni Karanka, ONS

  • Summary

    1. Before we start

    2. Defining & identifying statistical services

    3. Candidate services

    4. Developing statistical services

    5. Reusing statistical services

  • 1. Before we start

    • The identification, definition and adoption of statistical services is not an IT-only activity.

    • It relates to the capabilities the organisation wants to deploy: you need a mix of business leaders, statisticians and IT roles.

    • However, many of the tools and models that enable these activities come from an IT tradition (e.g., service orientation, business analysis, capability modelling, architecture).

  • 1. Before we start

    • You should be familiar and fluent with: – GSBPM – the statistical activity reference model

    – GSIM – the statistical activity information model

    – CSPA – the reference architecture for statistical production

    • We recommend you: – Have assessed your organisational maturity

    – Are aware of TOGAF, CSPA LIM, ESS EARF and SPRA

    http://www1.unece.org/stat/platform/display/metis/The+Generic+Statistical+Business+Process+Modelhttp://www1.unece.org/stat/platform/display/gsim/Generic+Statistical+Information+Modelhttp://www1.unece.org/stat/platform/display/CSPA/CSPA+v1.5https://www.opengroup.org/togaf/http://www1.unece.org/stat/platform/display/CSPA/CSPA+Logical+Information+Modelhttps://ec.europa.eu/eurostat/cros/content/ess-enterprise-architecture-reference-framework_enhttps://ec.europa.eu/eurostat/cros/content/spra_en

  • 1. Existing standards

    Contextual

    Conceptual

    Logical

    Physical

    Detail

    Level of

    abstraction

    What

    Specific Generic

    Business architecture

    Solution Architecture

    Application architecture

    Technological architecture

    Information architecture

    Statistical production reference

    architecture

    To be developed

    Business Capability

    Model

    GSBPM/ GAMSO

    GSIM

    CSPA V1.5

    CSPA LIM

    CSPA Architecture

    Patterns

    HLG Vision ESS Vision

    Architecture Building Blocks

    ESS framework

    UNECE framework

    To

    executio

    n

  • 1. Before we start

    • Ensure you have the right roles and channels:

    – Who in your NSI leads this work in the ESS?

    – Have your leaders committed to CSPA?

    – Do you have the people and resources that support the identification, sharing and/or implementation of statistical services?

    – Do you have the right collaboration tools and channels in order to work with other NSIs?

  • 2. Defining & identifying

    • A statistical service is a business service that applies to statistical production (it can be described using GSBPM and GSIM)

    • The statistical service allows access to a capability (organisation, people, process, technology)

    • Most reusable statistical services involve the sharing of the technology and process aspects of the capability, although a Service Provider can resource all aspects of a shared service

  • 2. Defining & identifying

    • We focus on sharing statistical services because:

    – They involve the greatest opportunity to realise savings and to embed best practice

    – We already have standards and frameworks that define them

    – Statistical production has many similar requirements and regulation drivers across NSIs

  • 2. Defining & identifying

    • The purpose is to identify services that:

    – Meet the needs of several NSIs

    – Can be developed to be reusable by the adoption of CSPA standards

    • To maximise these aims:

    – Follow SOA principles present in CSPA

    – Start from an understanding of business architecture and user needs, rather than available technology solutions

  • 2. Defining & identifying

    • Start from defining the service against GSBPM: – What function does it carry out? (GSBPM)

    – What information does it interact with? (GSIM)

    – Use this information to start populating a CSPA definition template

    • Note that the candidate services will be very generic at this stage and not include methods, technologies or non functional requirements

  • 2. Defining & identifying

    • Refine the definition:

    – What is the granularity, would it benefit from associated business functions?

    – How does it fit in the architecture of your NSI?

    • You can then compare your definition with those available (SCFE candidate service list)

  • 2. Defining & identifying

    • Example CSPA Definition

    • Editing methods not specified

    • Includes references to GSBPM and GSIM that can be extended in the Specification

    Name Determine Edits

    GSBPM 5.4 – Edit & Impute

    Business Function This service resolves missing or incomplete data using deterministic and selective editing rules

    Outcomes Missing or incomplete data is resolved

    Restrictions Applies to unit level data

    GSIM Inputs Dataset, Rules

    GSIM Outputs Dataset, Referential Metadata Set

    Service dependencies Requires set of editing rules, requires identifying incomplete data

  • 3. Candidate services

    • All candidate services need a complete CSPA Definition

    • This Definition can then be inherited by CSPA Specifications

    • ESS countries can use the CSPA Definition to assess investment, and agree donor and reusing organisations

  • 3. Candidate services

    • Now that candidate services have been defined using GSBPM and GSIM, you can:

    – Identify how they map against the ESS EA framework and the SPRA

    – Consult potential partners regarding user needs (preferably in the form of user stories)

    – Assess them using the AAA framework

  • 3. Candidate services

    Attractiveness

    qualifies and quantifies the benefit claims and assesses the contribution to strategic and operational objectives.

    Achievability

    evaluates the likelihood that the objectives can be achieved within stated financial, resource and timescale constraints.

    Affordability

    considers whether the implementation (develop, reuse) costs , relative to its realizable benefits,

    are reasonable.

    AAA Assessment Framework

  • 3. Candidate services

    • We favour a multi-criteria analysis of benefits: sub-criteria for each A and agreed weights for each sub-criteria

    • Benefits of each project / service is described with scores used to prioritise

  • 3. Candidate services

    • Once candidate services have been prioritised, one of several actions can ensue:

    1. The candidate service is to be developed

    2. The candidate service resembles an existing solution, that can be refactored or developed

    3. The candidate service is seen as requiring further decomposition

    4. The candidate service is not considered suitable for development (e.g., no benefits or users)

  • 4. Developing statistical services

    • To develop statistical services, the service should have:

    – A clear business benefit (assessed by AAA)

    – Identified potential ESS users (“CSPA investors”)

    – Documented business or user needs (preferably in the form of user stories)

    – Consideration of relevant legislation (e.g., security and information assurance, data sharing, software licensing)

  • 4. Developing statistical services Investor role Service Designer role Service Builder role

    “The investor role is focused on directing and financing the investments in new statistical processes that are needed to produce new statistical products.”

    “The Designer assesses what solutions (services) are available to actually realize this design… If no suitable existing solution can be found the Designer will specify the functionality for each Statistical Service needed to meet requirements.”

    “The Service Builder receives a Statistical Service Definition and a Statistical Service Specification from the Designer. The Service Builder then implements the Service.”

    Provides: • User needs • Requirements • Priorities

    Provides: • Service definition (joint with investors) • Service specification

    Provides: • Service implementation

    Likely to have many ESS investors for each service.

    Often is one of the investors.

    Works closely with the Service Designer. Distinct from Service Provider.

  • 4. Developing statistical services

    • The next stage of the development is working on the CSPA Specification of the service: – What methods / features / requirements will it

    have? How do user stories relate to these?

    – What logical information objects does it use (here CSPA-LIM extends the GSIM objects)?

    – What is the default approach to sharing the service?

    – What are the likely architectural patterns?

    – How will the Service be accessed / used?

  • 4. Developing statistical services

    • There are three approaches to service reuse:

    – Interoperable. The organisation develops and operates its own service which can exchange information and interact with other solutions.

    – Replicated. The same service is operated in more than one organisation (duplicated).

    – Shared. A single instance of the service is operated centrally by one organisation and used by many organisations.

  • Example: ESS 2020 Vision

    Sharing approaches for ESS 2020 Vision building blocks

    One size does not fit all: • Inconvenient to share common Process Orchestrator • Different national legislation likely to cover IT Security & Identity Management

    Interoperable

  • 4. Developing statistical services

    • There are many models of building the statistical service (in-house, outsourced, etc.)

    • In all cases ensure:

    – the application architecture of the service is traceable to the Specification

    – the delivery team follows the application, information and technology principles of CSPA

    – investor needs / requirements are available

    – CSPA Definition and Specification are kept up to date (as development is likely to be iterative)

  • 5. Reusing statistical services

    • The first step to reusing a statistical service is finding out whether a suitable candidate exists in the Service Catalogue

    • Three outcomes are possible:

    – One exists that meets the requirements

    – One exists that can be enriched to meet the requirements

    – No statistical service exists that could meet the requirements

  • 5. Reusing statistical services

    • If no statistical service exists, the process of defining the service should start

    • In this scenario, the organisation that needs the service is likely to become a Service Designer and/or an Investor

  • 5. Reusing statistical services

    • If a statistical service that can be enriched is identified, the organisation that needs the service should contact the Service Provider

    • The Service Provider and Service Builder should assess whether:

    – The service can be enriched

    – A new service has to be built

  • 5. Reusing statistical services

    • If a statistical service exists that can be used, the organisation should assess how they can assemble the service

    • The approach to sharing the service becomes key for the Service Assembler in the receiving organisation

  • 5. Reusing statistical services

    • If the service is to be replicated:

    – How will it interact / coexist with other services in the NSI? How will it be orchestrated / choreographed?

    – Does an environment need to be set up for the service? How will it use information?

    – Are there other barriers such as language support, lack of maturity in the architecture of the receiving NSI (e.g., lack of alignment with GSBPM, no SDMX or DDI data standards)?

  • 5. Reusing statistical services

    • If the service is to be shared: – What is the protocol to access the service? What

    steps are needed to access the service?

    – How will it interact / coexist with other services in the NSI? How will it be orchestrated / choreographed?

    – Do the NSI services & data meet standards required by the shared service?

    – Does the shared service meet performance & security NFRs when accessed remotely?

  • 5. Reusing statistical servivces

    • If the service is to be interoperable:

    – Consider the development needed to build a new implementation of the service

    – Are the specifications of the existing service reusable or easily amended?

    – Can any of the technology components of the available service be reused?

    – Ensure all technology & information requirements for interoperability are met

  • To Conclude

    • This guidance is under development and being extended based on the outcomes of the ESSNet on Sharing Common Functionalities in the ESS, for example:

    – Ways to disseminate / maintain the guidance

    – Role of a Centre for Excellence

    – Experiences of service implementation

    – How to update the service catalogue

    – Best practice and models for service assessment

  • To Conclude

    • If you have any comments or feedback, please contact: [email protected]