Each project arrives with its purpose, milestones and first tasks written out. Here is how the first three read.
Workload requirements sheet before provisioning
Teams that open the cloud console before writing anything down end up sizing for a guess and paying for it twice. One page listing expected users, requests per second at peak, data volume after a year, latency targets, uptime promise, recovery point and recovery time, and any data location rules gives every later choice in this area something to be measured against.
First tasks
- Write down expected peak requests per second and active users for the first year
- Estimate stored data after twelve months, including files and backups
- Agree recovery point and recovery time targets with the product owner
- List every data location or compliance rule that customers or regulators impose
Choosing a cloud provider for the first workload
Switching providers after launch costs months, so the choice deserves a week of evidence rather than whichever console someone used at a previous job. Scoring two or three providers against the requirements sheet, the team's existing skills, regional coverage, managed database options, support plans and startup credits turns a habit into a decision you can defend to investors and new hires.
First tasks
- List the managed services the requirements sheet needs from any provider
- Score two or three providers on services, regions, support, cost and team skills
- Deploy a hello-world service on the front runner to test the tooling
- Write a one-page decision note with the scores and the reasons
Separate accounts for production, staging and sandboxes
Running everything in one account means a mistyped command in a test can delete production data, and nobody can see which environment is costing what. Splitting into separate accounts or projects for production, staging, shared services and personal sandboxes, grouped under one organisation with consolidated billing, gives each environment its own blast radius and its own bill.
First tasks
- Sketch the account layout for production, staging, shared tooling and sandboxes
- Create the organisation and turn on consolidated billing
- Create each account and connect them to single sign-on
- Draw the final structure and add it to the infrastructure README