openIDL - Target Architecture

High Level Requirements of the System

Notes from the team

KS

  • support stat reporting
  • support ad-hoc data calls
  • verify data availability
  • Data stays private to the carrier
  • Only results of extractions leave the carrier
  • common extraction request across all nodes
  • common data model for extraction across all nodes
  • Any one extraction uses the same model for all data owners
    • JM - Agreed, but per level of the published model
  • trust extractions - we are executing code after all
  • Correlated data can be accessed as part of the extraction
  • All updates to the system are well managed
  • Support multiple "footprints"
  • physical db schema maintenance is minimized
  • Technical choices for implementation can vary from carrier to carrier for those items that reside in the carriers perimeter
  • Passes audit by All members of TSC

JM

  • Security model has white hacks as part of regression testing
  • Done when everything is in a comprehensive regression test base and all tests pass
  • Each major box has "push button" install process
  • Reference tables all pre-populate as part of HDS install
  • DDL in the db to build out the model in each major technology
  • Test records self install to HDS and test base runs.
  • Capacity and DR specifications are published and tested

TE

  • Insurer needs a standard for regular Policy-level "experience recording" assertion
  • Insurer's data moves from batch/chunk integrity at load, to Per-Policy integrity over time and at time of inquiry
  • Analytics node is the "box" when we talk "openIDL in a box" as it determines the value of the information as a result (who gets the analysis and why) - we need different types/sizes as well as Orgs/Roles (todal it's all AAIS)
  • Analytics Node host ("information seeker") or Seeker's Agent (e.g. NAIC, PCI, etc. on behalf of >1 Seekers) for Org/Roles/purpose - creating Extraaction Patterns, etc. (today is AAIS or whoever deploys the network - ND, MS, etc.)
  • "AAIS" cloud/node(s) need to become "(Stat) Agent" orgs/nodes (>1) acting on behalf of >1 Data Owners

SB

  • There is a test-net and a main-net
  • Governance Framework required for why (principles) AND how (mechanics) things get done
  • States and prospective members can Pilot/POC via testnet?
  • openIDL is running at least 2 nodes (CA and analytics) to operate NOC-like service
  • Is there "one node architecture" for all or would a state have a different kind of node (please say no)
  • Applications (stat reporting, etc)
----
openIDL Network
----
openIDL TestNet


Target Application Architecture

Details of the  Member

Discussion about the adapter

Target Network Architecture

Target Data Architecture

See Technical Considerations

Target Technical Architecture

Digging into the integration between the hosted node and the carrier.


Feedback on Current Architecture and Implementation

See this site for feedback from Travelers based on deployment experience with the current architecture.