Planning Digital Ocean Environments: Production and Sandboxes

Major points:

• No redundancy

Structure of the projects envs:

Production env(old architecture) :

2 Load balancers → routes to 2 droplets → Droplets are Linux VMs → The main droplet runs the container config from the docker-compose file → The second droplet is responsible for Sendgrid → 2 database clusters 1 for Redis and 1 for PostgreSQL

Staging env (new architecture):

1 Load Balancer → 6 Droplets →

Decisions:

• Docker-compose – DevOps

• Database-managed instances (PostgreSQL) – DevOps + Devs for connection

• Redis managed instances – DevOps + Devs for connection

• Every service needs to have at least two instances – Devs

• Decide Sendgrid solution (setup) – Devs

• Prevent sending of emails before production is approved. – Devs

• Be sure not to load the testing files in the production bucket. – Devs

• Do not load the Storybook container to the production env.

• We do not need the separate BI board anymore cause it is a part of the new architecture. – DevOps

• Create a new project on Sentry and connect it to the new Architecture. – Devs after decision

• Anton and Yurii will decide the exact solution for Logging and Error handling
  (Senty \ NewRelic etc.) = Anton, Yurii (Devs)

• Twilio for SMS services. Be sure to not send anything before live. Investigate the env and accounts. – Devs

• Look for secrets management for production. Decrease the list of people who can access it. – DevOps

• Write a new CI\CD pipeline for Production env. – Devs + DevOps

• To come up with the right resource plan for the Production env