Google Cloud architecture

Google Cloud diagrams hang off the project boundary, and services within a project are usually reached through well-known managed endpoints rather than through network topology.

That makes text diagrams work particularly well here, because there is less positional meaning to preserve.

#When to reach for one

• Design documents, especially for serverless and data pipelines.

• Explaining a Pub/Sub topology and who subscribes to what.

• Cost and quota conversations, where the service list is the subject.

#How to draw it in Capable

#A worked example

A Cloud Run and Pub/Sub pipeline:

The source, which you can paste into a new diagram and edit:

flowchart LR
  U([Users]) --> LB[Cloud Load Balancing]
  LB --> RUN[Cloud Run: api]
  RUN --> SQL[(Cloud SQL)]
  RUN --> TOPIC{{Pub/Sub: events}}
  TOPIC --> WORKER[Cloud Run: worker]
  TOPIC --> BQ[(BigQuery)]
  WORKER --> GCS[(Cloud Storage)]
  WORKER --> SQL

#A few things that catch people out

• Draw the project boundary; IAM and quota both follow it.

• Pub/Sub topics deserve to be nodes. Hiding them loses the fan-out.

• Regional versus multi-regional is worth annotating if latency or cost is under discussion.

#Related

Draw the trust boundary.