Decouple Stage/Production [48 hours]

[rhodges]

For this to work we will need:

• A new Server
• A process to migrate layer data
  • Ideally only NEW and updated layers
• A backup/restore system

[rhodges]

Option 2:
Keep the "shared" database, but separate the code bases.

How this tackles the issues:

• Application code can advance on Sandbox and Stage without impacting production
• Stack updates (below Python) will be tested on Sandbox, mitigating need to test again on stage
• Database updates will be tested on Sandbox, mitigating the need to test again on stage
  • This assumption should be reviewed
• Performance can be improved by boosting specs on single server, probably cheaper than spinning up a second server of same power
• pushing new layers to production could be done without risk from either Stage or Prod sites

[rhodges]

Problems with option 1:

• High risk of data loss, ID mismatch, missing dependencies when performing export/import
  • OR: no data would be allowed to be added, edited, or removed from production (BAD!)

Problems with option 2:

• Bug resulting in HW failure for Staging would take down Production
  • Staging could be served from a separate server and connect to the production DB
• Updating/Migrating the Stage DB would also result in migrating the production DB, possibly leading to errors