Page History

...

• Sean Bohan (openIDL)
• Mason Wagoner (AAIS)
• Joseph Nibert (AAIS)
• Dale Harris (Travelers)
• Ken Sayers (AAIS)
• Yanko Zhelyazkov (Senofi)
• Satish Kasala (Travelers)
• Peter Antley (AAIS)
• Faheem Zakaria (Hanover)
• Tsvetan Georgiev (Senofi)
• Ash Naik (AAIS)
• Dhruv Bhatt (Hartford)

Agenda:

• Update on ND POC
• Update on openIDL Testnet (Jeff Braswell)
• Update on internal Stat Reporting with openIDL (Peter Antley)
• Update on Infrastructure Working Group (Sean Bohan)
• MS Hurricane Zeta POC Architecture Discussion (KenS) 
• Architecture Decision Capture Process (KenS) - Discussion
• HDS architecture decisions (PeterA) - Discussion
  • Schema Management via Excel vs SQL and Flyway
  • Database vs Data warehouse
  • Expiration dates on codes
  • mutable vs immutable codes, (why would we recycle a code)
• openIDL contributor engagement (James Madison) - Discussion
• AOB:
• Future Topics:

...

Notes: (Notes taken live in Requirements document)

• ND POC (KenS)
  • Second data call today
  • All carriers have data in 
  • Second report for VINs not registered
•  openIDL Testnet (JeffB)
  • pickup in action: attempts to get carriers on board
  • coord and orchestration
  • contributor engagement
  • Hanover - map data to auto stat plan format
  • Travelers & Hartford
  • coming weeks and months: significant progress
  • work together on mult fronts
• Internal stat reporting (PeterA)
  • decode tables, artifacts, schema management
  • updates - JamesM requested a detailed presentation on data layer - March 27
  • how can we offload some of the work onto business poeple and changing dimensions over time if at all
  • phys join of tables, claim and policy views, modeling of steps
  • expanding documentation significantly
  • every other month since june is settling into mature
  • decode -stat plan and personal auto - foundation done this AM
  • carriers - need to put stat data into HDS, produce stat records per stat handbook
  • only need to load stat records and can start reporting
  • (peter screenshare)
  • https://semver.org/
    https://github.com/openidl-org/openidl-main/tree/etl-pka-flyway/openidl-hds-ddl/ref/pa/tables
  • https://github.com/openidl-org/openidl-main/blob/etl-pka-flyway/openidl-hds-ddl/ref/pa/tables/V0.0.1.6__pa_stat_stg.sql
  • utilizing JS to write out sql, sql scripts managed by flyway
  • work with proper keys for joins
  • Dale - James thought we might put results in excel so excel tab for each table so tables create definitions
  • documentation - practice of updating data model
  • coverage codes - json, states, can upgrade json from excel, converted from json to excel, config doc + js file
  • tools - bolt ons on top of db
• Joseph: FlyWay
  • <demo>
  • containerization from dev side
  • FlyWay - great db management tool, as a container
  • FlyWay has versioning schematic - they went with semantic version
  • maintains version control
  • containers will make it easier
  • operationally what the data model will drive
  • educating carriers to do this
  • ran earlier, can stop it, migrates all of these
  • can do it in source control
  • Flyway effectively does checksum on every file, maintains data integrity
  • parallel for EP schema? 
  • EP written by dev, could set up repo and put in flyway to keep EPs, but each EP should be kept in Git somewhere, not re-ran, on ledger best ref point
  • 2 parts - creation of EP schema, other is
• Mason analytics
  • peter task: write simple sql tables for lines of business
  • get idea of where we are, what completed, map out getting tables up and running will look like
  • personal auto, went thru, 33 tables needed to bring over to use
  • GT2
  • most tables from statplan directly
  • some stat plans have mult lines of business
  • some repeatability in categories, found things in stat plan, not which need table to be integrated
  • more breakdown of how each stat plan is looking to keep account of
  • 201 tables out of all stat plans 
  • completed 99 of 201
  • view of how this is looking for timing and expectations
  • code tables - take codes and give description
  • for each code that comes in we need a ref table, and we need a way to decode, load, make script for each table and view, put in flyway, allow easy managed db across all these areas
  • codes across diff business lines the same? very few to none
  • none can be skipped w/o thorough review
  • coverage codes - lot of copy-paste
  • will speed up a lot as they get going, not as much overlap as you think
  • transaction code would have secondary key as line of biz? 
  • opportunity to do analysis across lines of business
  • look out for it in a few places, def opportunity for cross-lines
  • have to get to next level beyond what Mason showed - data dictionary for all of them
  • intern request
  • can do it for tables named the same across lines, see where they lined up
  • version zero - can change anything we want and what a good releaser looks like
  • goal for this week - get auto coverage report gen again based on high end ETL decode job
• Architecture Decision Capture Process (KenS) - Discussion
  • Process for capturing decisions
  • openIDL - Architecture Decisions

Recording: