[SPIKE] s3 lifecycle rules for backups

[BasilMawejje]

PT Story:

PT URL: https://www.pivotaltracker.com/story/show/173667123

Changes proposed in this pull request:

1. Spike using aws sdk to transition backups in s3 to different storage classes e.g transition from S3 Standard to Standard IA after 30 days, then transition from standard IA to Glacier after 90 days and finally transition from Glacier to Deep Archive after 180 days.
2. Spike also sets expirations for file storage in S3. For example we can explicitly set a rule to expire objects after 365 days.
3. Requirements will be discussed further to make sense of this spike. I made a number of assumptions while I wrote the lifecycle method.

Screenshots (Optional):

N/A

Ready for review:

@

Ps:

• No tests added because I don't have the .env file to setup the test database.

[weedySeaDragon]

I put up a wiki page that has the backups schedules: https://github.com/AgileVentures/shf-project/wiki/Backups-on-AWS

Other Misc. thoughts:

• will want to read the configuration from a file (ex: number of days to keep for different types of backups, storage class). Perhaps this should be the XML that AWS uses (ex: those shown on https://docs.aws.amazon.com/AmazonS3/latest/userguide/lifecycle-configuration-examples.html)
• default zone should be Stockholm
• should we also be setting lifecycle rules on buckets? (Would this be easier? Should we do this instead of setting lifecycles rules on files or in addition to?
• do we want to keep backups in 2 different zones? If so, which backups and which additional zone? (maybe monthly and yearly? maybe additional zone should be in the US?) what will the cost be?