From Enterprise Strategy to Delivery – Architecture that delivers in Financial Services
eArchitecture » Page 'The Domain specific Operational Data Store'

The Domain specific Operational Data Store

  1. Creating a single operational store for an enterprise is a titanic endeavour for which a business case would be hard to justify. More pragmatically, domain Operational Stores, operating in loose federation, brought together by a light-weight services layer, in an MDM pattern, probably via an ESB, is possible. A unifying information model, joining the domain (IFW, FSLD, etc) will simplify the creation of the services, events layer, reducing the workload on the ESB. Not a new idea, proposed by Garner in the 2000s. Important components: data ingestion technologies (data replication, ETL, ESB, feeds), enquiry services and events (views, services, events, ESB), real-time rules engines for operational value (AML, fraud, operational rules, etc), connectivity and access.
  2. What is it?
  3. Where is it positioned? 

    The ODS boundaries need to be carefully defined to separate it from a Data Warehouse or System Data Store.
  4. ODS Patterns

    The ODS will be a critical part of the information network of the enterprise, by design it will be customer impacting, so the design and operational aspects are critical and must be balanced.
    Data synchronisation, Availability, Performance.
  5. Interesting the relationship between Data and Event Management (MDM and CEP) comes together in the ODS.
  6. ODS Technology

    The technology solution has to cover ingestion, storage (days and months), distribution (enquiries and events) in real-time, at data scale and data complexity (multiple sources, formats, management  (change over time and retention)). Storage in the ODS will need to be canonical (domain standard format) and in Raw system format. Distribution and access are key to creating value from the ODS, enabling both the originating and new systems access to the data and generating events enable a proactivity enterprise. Events based on data require a Rules Engine capable of processing ODS data store intelligently, i.e. AML/fraud engines, reconciliation, etc.
  7. Does the three views (Traditional, Stream, Scale) presented by IBM miss out the Operational Data Store?
  8. Oracle handle the data ingestion
  9. Workspaces in 11g
  10. Oracle Data Integrator (ODI), business rules application to data 
Posted in Technology

Leave a comment

Top of page | Subscribe to new Entries (RSS) | Subscribe to Comments (RSS)