Skip to content

Latest commit

 

History

History
51 lines (31 loc) · 2.18 KB

0003-initial-architecture.md

File metadata and controls

51 lines (31 loc) · 2.18 KB

3. Initial system architecture

Date: 2022-03-30

Status

Accepted

Context

This is a greenfield build project, to evaluate options and settle on an initial architecture we have:

  • Defined the functional scope of a MVP towards which we're building
  • Mapped our understanding of the systems surrounding the MVP
  • Surveyed and reviewed existing TTS projects and documentation such as https://engineering.18f.gov/
  • Reviewed and discussed proposals as an engineering team

Proposed architecture which was reviewed by the engineering team: Initial architecture

Decision

Finding no immediate needs not met by TTS default choices for infrastructure and datastore, we will build on a foundation of cloud.gov and;

  • Use cloud.gov as our PaaS for development, staging, and production environments
  • Use cloud.gov brokered Postgres AWS RDS instances for all data
  • Use cloud.gov brokered S3 instances for storage of PDF blobs.

The frontend will be developed and hosted on Cloud.gov Pages. We will maintain separation between the frontend and backend as distinct systems. This separation will;

  • Allow us to leverage Pages functionality to streamline prototyping and development efforts
  • De-couple development from the backend. A contributor to the frontend need not have any knowledge of the code, infrastructure, or build pipeline of the backend.
  • Not add additional risk or cost should we later decide to serve the frontend from the backend system.
  • Serve as the first consumer of the backend API

We will use Python & Django for the backend. Using Django will allow us to:

  • Move quickly given the current engineering team's background
  • Leverage existing patterns from other TTS efforts such as civilrights.justice.gov
  • Provide functionality such as the admin interface which we believe will fulfill some MVP requirements with minimal modification

Consequences

  • We'll establish separate development and build environments for the backend and frontend
  • The backend will be written in Python and we'll strive for idiomatic Django implementations leveraging existing work where ever possible.
  • We'll need to tackle a legacy data migration from Oracle -> Postgres