Migration to Redis

[kzalys]

After analyzing the use case of this service more I realized that Redis would suit hs_auth better than MongoDB as the data storage technology.

Why Redis?

• Performance. Redis is an in-memory key value store which means it is much faster than MongoDB which stores data on the disk. Considering the fact that auth methods on hs_auth are called pretty much whenever any call is made to any hs service, improvements in response time should lead to the performance of the entire system and, as a result, a better user experience. Though the real performance impact is yet to be determined and I will be adding more information to this issue once I've performed some tests.
• Persistence. In-memory datastores usually only store the data in memory which means that when the datastore process is killed, all data is lost. Redis is different to other in-memory datastores by supporting on-disk persistence meaning that even if the host goes down, the data will not be lost.

Things to note:

• Redis uses the memory rather than the disk to store data which means that a large amount of memory would be required if the amount of data to store gets large. However, in the case of hs_auth each user can create at most 2 objects on the database (the user object and a team object). Considering we will most likely never have more than 300 users, the amount of memory required to store all of hs_auth's data would be minimal.

Things to do:

• Implement an interface for storing user and team objects on Redis
• Update the docker setup deploy Redis instead of MongoDB
• Allow the developer to choose wether to use Redis or MongoDB (low priority)