ADR Generator from Notes

Who it's for

• Software architects documenting design decisions
• Technical leads writing meeting notes
• Engineering managers reviewing project trade-offs
• DevOps teams maintaining asynchronous documentation

Use cases

• Documenting a new database schema after a team meeting
• Recording the reasoning behind a recent infrastructure change
• Creating a knowledge base for onboarding new team members
• Justifying a technical debt payoff strategy to stakeholders
• Preserving institutional knowledge during team turnover

FAQ

Sample output

**Title:** Replace Postgres with SQLite for Billing Service
**Status:** Proposed

**Context:** The billing service requires a database to store and manage billing information. Initially, we considered using Postgres (Aurora) or MySQL (PlanetScale) due to their high scalability and performance. However, after re-evaluating our requirements, we are considering using SQLite (Turbo) as a simpler and more lightweight alternative.

**Decision:** We will use SQLite (Turbo) instead of Postgres for the billing service. This decision is based on the fact that our billing service will have a limited number of writers, with only one writer per billing row, which reduces the concern of write contention.

**Consequences:**
* Positive: Reduced complexity, easier setup and maintenance, and lower operational costs.
* Negative: Write contention may still occur at high write volumes (>200 writes/sec on a single billing row), potentially leading to performance issues. Additionally, SQLite may not scale as well as Postgres or MySQL for very large datasets.

**Alternatives considered:**
* Postgres (Aurora): Rejected due to increased complexity and higher operational costs. While Postgres provides high scalability and performance, it may be overkill for our current billing service requirements.
* MySQL (PlanetScale): Rejected due to similar concerns as Postgres, including higher operational costs and increased complexity. Additionally, MySQL may require more maintenance and tuning to achieve optimal performance.

Mitigation strategies for the potential write contention issue include sharding by tenant in phase 2, which will help distribute the write load and reduce contention. We will monitor the performance of the billing service and adjust our strategy as needed to ensure optimal performance and scalability.