Architecture
A regulated GKE platform, multilingual analytics, the foundation, and the services by job.
A regulated platform on GKE, run through pipelines
A reference design for regulated workloads: every project and every release comes from code.
- No manual deploymentsEvery release and infrastructure change through CI/CD, with security and quality gates.
- Projects from codeEvery project, in every environment, created and changed by Terraform.
- Microservices on GKEReleased with minimal downtime.
- Edge to containerEdge security, container scanning and a SIEM, watched every day.
Point at a component to light its links; select it for what it does here and the practice that makes it right on Google Cloud. A Google Cloud product without an icon of its own shows its category's icon, as Google's icon set does.
Multilingual review analytics on Google Cloud
A reference design: daily pipelines, two models, a quality checkpoint and dashboards the next day.
Point at a component to light its links; select it for what it does here and the practice that makes it right on Google Cloud. A Google Cloud product without an icon of its own shows its category's icon, as Google's icon set does.
A Google Cloud foundation, layer by layer
What NGT puts in place before the first workload lands.
- 01Resource hierarchyOrganization, folders per environment, a project per service and environment.OrganizationFoldersProject factory
- 02Organization policyLocations, external IPs and allowed services set once, everywhere.Organization Policy
- 03NetworkOne hub for every VPC, a private link to the data centre.Shared VPCNetwork Connectivity CenterCloud Interconnect
- 04IdentityOne sign-in federated with the customer's directory; roles to groups.Cloud IdentityIAM
- 05Logging and securityAudit logs from every project in one place; findings in one view.Cloud Logging sinksSecurity Command CenterCloud KMS
Google Cloud services NGT uses, by job
| Job | Google Cloud services |
|---|---|
| Move servers | Migrate to Virtual Machines |
| Move databases | Database Migration Service, Cloud SQL |
| Run containers | GKE |
| Move and shape data | Dataflow, Pub/Sub, Dataproc, Cloud Composer |
| Analyse | BigQuery, Looker |
| Build models | Vertex AI |
| Operate | Cloud Monitoring, Cloud Logging |
| Build and provision | Terraform, Cloud Build |
Questions these designs answer
The questions this part raises, and their answers.
Every project is created by Terraform through the pipeline, from shared modules. Nothing is changed in the console, so drift has nowhere to start.