Prepare your own cloud
Plan account access, networking, capacity and a deployment review.
Understand the deployment boundary
In a bring-your-own-cloud design, your cloud account hosts the event-processing resources. A control plane coordinates configuration and operations.
| Your cloud | Platform coordination |
|---|---|
| Source and sink connections | Deployment configuration |
| Event processing and storage | Operational status and health |
| Networking and data residency | Workspace and access workflow |
Prepare the account and region
- Choose AWS, Google Cloud or Azure and the region where the data should live.
- Identify who owns the account, billing, networking and access approvals.
- Document source and destination endpoints and their access constraints.
- Review any residency or retention requirements with your team.
Avoid copying a reference product’s account IDs, subscription links or provisioning commands into your deployment.
Review provider-specific resources
| Provider | Account scope | Cluster planning | Access review |
|---|---|---|---|
| AWS | Account and region | EKS, storage and VPC networking | IAM roles and scoped resource permissions |
| Google Cloud | Project and region | GKE, storage and VPC networking | Service accounts and IAM permissions |
| Azure | Subscription and resource group | AKS, storage and virtual networking | Managed identities and role assignments |
Choose resource sizes from your workload and availability requirements. A reference deployment size is not a substitute for your own capacity review.
Plan networking and secrets
- Confirm DNS resolution and routes to your sources and destinations.
- Choose private connectivity where required.
- Review management traffic separately from event data traffic.
- Use a cloud secret manager for connector credentials.
- Confirm encryption settings for transport and storage.
The security page covers workspace roles and the questions to bring to a review.
Estimate the workload
Start with average and peak write throughput, retention days, consumer groups and processing complexity. Replication, compression and destination reads also affect capacity.
Use the workload planner to estimate daily data volume and raw retained storage. It does not calculate a cloud invoice.
Average write throughput: 10 MB/s
Daily incoming data: 864 GB
Retention: 3 days
Raw retained data: 2.592 TB
Consumer groups: 3
Combined read rate: 30 MB/sThese decimal-unit estimates exclude replication, compression, indexing and other infrastructure overhead.
Check readiness before production
- Validate a small pipeline.
Run the sample-event tests from the pipeline guide.
- Review access boundaries.
Confirm the workspace roles, cloud permissions and credentials that will be used.
- Test interruption and recovery.
Check source disconnects, destination failures and the intended retry or replay process.
- Assign operational ownership.
Name the people responsible for alerts, capacity and deployment changes.
- Review deployment scope.
Document the selected region, resources, estimate and rollback plan before provisioning.