Google Cloud · BigQuery · Diagram as code
BigQuery architecture diagram
Show how events and batch jobs land in BigQuery. ArchLex checks writes and reads from Dataflow, Cloud Functions and Cloud Run against the GCP catalog.
Dataflow from Cloud Storage into BigQuery.
Batch ETL: read a lake bucket, write the warehouse.
provider gcp
direction LR
job: dataflow["ETL job"]
lake: cloud-storage["Raw data"]
wh: bigquery["Warehouse"]
job -[reads]-> lake
job -[writes]-> whPub/Sub to Cloud Functions to BigQuery.
Lightweight streaming path without Dataflow.
provider gcp
direction LR
events: pubsub["Events"]
fn: cloud-functions["Loader"]
wh: bigquery["Analytics"]
events -[invokes]-> fn
fn -[writes]-> whPattern notes for BigQuery.
Placement and relationship rules ArchLex enforces for this service.
- Placement. BigQuery is a regional analytics service—keep it at the root or under account/region, not inside a VPC subnet.
- Valid connections. cloud-functions|cloud-run|compute-engine|dataflow -[writes]-> bigquery and the same sources can -[reads]-> bigquery. Pub/Sub often sits upstream and -[invokes]-> a function that writes the warehouse.
- Warehouse patterns. For streaming ETL, prefer pubsub → dataflow/cloud-functions → bigquery. Label edges when the same warehouse serves Looker or other BI tools downstream.
Ask an agent to draw it.
Install the ArchLex skill or connect the free MCP server, then paste this prompt.
npx skills add baires/archlexDraw a BigQuery architecture diagram with Pub/Sub invoking Cloud Functions that write to BigQuery, and a Dataflow job reading Cloud Storage into BigQuery. Validate it and give me a playground link.Prefer MCP? Follow the MCP server guide.
Related Google Cloud examples.
BigQuery diagram questions.
How do I diagram BigQuery pipelines in ArchLex?
Use the bigquery resource as a writes or reads target from dataflow, cloud-functions, cloud-run or compute-engine. Put pubsub upstream when you need an event bus.
Can I show Looker or Dataplex next to BigQuery?
Yes when those catalog entries exist in your ArchLex version. The gallery includes a Data Warehouse example with Looker and Dataplex.
Does ArchLex know about datasets and tables?
No. BigQuery is one node per diagram resource. Encode dataset names in the display label if you need them visible.
Is BigQuery allowed inside a VPC?
No. Keep analytics services outside subnet scopes unless you are drawing a specific networking accessory as its own resource.
More Google Cloud service diagrams.
Open source · MIT licensed