Establish a strategy to shard off DB load balancing from the main redis instance

As noted on &877 (comment 1252967626), to start work on this effort we need to establish (at least) the following points of our plan:

• What is our tolerance for data loss on this feature, and how does that impact our data migration strategy?
• Is the redis.yml configuration file being introduced on &886 apt for this effort?
• Should we consider provisioning this cluster on k8s?