Six example models next to the three C4 ones: arc42 (Cinema Tickets, IoT Telemetry, Team Wiki) by section, ArchiMate (Insurance Claims, HR Onboarding, Logistics Tracking) by layer, each validated with 0 errors and 0 warnings. resolveRefs and the editor schemas now take ref: in workflow and dataflow nodes, sequence participants and lifecycle states, as the skill already did. Example tests check each framework's folders.
21 lines
772 B
Plaintext
21 lines
772 B
Plaintext
schema_version: 1
|
|
decision:
|
|
id: timescaledb-for-telemetry
|
|
title: TimescaleDB for telemetry
|
|
status: accepted
|
|
date: 2026-09-03
|
|
elements: [tsdb, ingest, ops-dashboard]
|
|
diagrams: [0500-building-blocks/level-1.archify, 0700-deployment/production.archify]
|
|
context: >
|
|
The fleet sends about 4,000 readings a second. A plain
|
|
PostgreSQL table grew too fast and dashboards slowed down
|
|
after a few weeks of data.
|
|
decision: >
|
|
Store telemetry in TimescaleDB hypertables with hourly
|
|
continuous aggregates. Raw data is kept 90 days, rollups
|
|
10 years. The team already runs PostgreSQL.
|
|
consequences: >
|
|
SQL stays the query language for the API and the
|
|
dashboard. Retention jobs must be monitored. Very high
|
|
device counts may need sharding later (see risk R1).
|