Infrastructure Working Group - this week, Thurs 12 at noon ET
Next AWG Call - Monday Sept 11, 2023 (NO CALL NEXT MON DUE TO LABOR DAY)
Next Technical Steering Committee Call - Thurs Sept 7, at 12 noon ET
B. HDS and carrier side:
How will carriers load the HDS
How will carriers review data call requests
How will extraction patterns be executed
Possible containerization of the extraction processor (?)
Notes:
PA
OLGA status
currently building OLGA
good solution for all at AWS
need some porting to run azure
not sure where with their timelines
OLGA is good for taking a large set of multi-line stat data, ingesting it
rapid way
then, doing large amt of validations quickly, allowing users to update single and bulk records
for Cat reporting, a lot more sparsely populated
way less attributes for stat record
not sure if loading the cat reporting may be more trad load, need to come up with plan/timeline for that
whoever is operating that node has PostgresDBm load stat records into DB, has scripts to do that
maybe first iteration could be manual
every carriers has HDS
stat record is positional record
AAIS has one script, runs mapping, maps stat record to json, another script makes insert tables and inserts all into DB
doesnt take long, some operator runs on local machine
possible to extract data for CAT POCs, put into format to help loading
basically stages of data processing, relatively doable
D or F? reasonable way to do POCs with managed data loading
F - great way to get folks on network, still need a true node somewhere
node as povided for connection, talking about just loading data into hds
in case of azure, native, try to have native environs for HDS to manage data, alt - prep data for trusted provider
part of discussion, architecturally, you will have HDS
immediate POC discussion might be spin off
good etl tools, might be easier to put directly in file db
outputting format not a big issues, getting it together, and aligning it with ea column in schema - is it agg, derived, inc cert subset - getting THEN formatting THEN calculating prooperly is the challenge
same table used for all, same columns
attributes we discussed
Evolution of the data model itself, not just policy and claim event records
struct of the db, what qualifies
combination of data modeling and assets.
SDMA and legacy systems, keel two tables together
during load process, all together
filter based on transax code
2 tables in DB
ingested, call in same table
How data call requests would be reviewed
when you consent to data call it is executed
provision, requested, can consent to data call but executd at a later date
Roadmap?
RIght now - consent = execution, open item for deferred execution
part of UI to review data calls, ability with node-as-a-service, have UI interrogate request, download, examine and try locally
running sql to local table before you commit and direct to response
how to host UI, interact with db
orchestration and control to be housed closer to db
done at the carrier side near HDS
Extraction
description of what was requested goes with the executable
when it does execute, run against EP schema
executing SQL, can't modify HDS
EP Schema with Peter's POC
EP Schema temporary
ep permission to create temp tables, after EP run that would be wiped clean
expecting on EP running at a time in the machine
detail, phys workflow
Containerizing
R&D topic
need some more insight into the diff EPs and data calls
wont be able to address, how we have more than 1 pattern
stat reporting was so complicated - may be worth having it as an app container to do the report
talking about stat reporting lately
some agg fields
Time
Item
Who
Notes
Documentation:
Notes: (Notes taken live in Requirements document)