In looking at your intended dataset objectively, what would happen if it was lost? If your tenant application is simply caching data that is helpful, but not necessarily mission-critical, then Redis might be a good solution. Or if you can reload data for your Redis instance from another source of truth system quickly, then Redis may also be a good fit for your application.
For example, if your application is simply using Redis as an in-memory session cache for a website, you can probably disable snapshotting and AOF for maximum performance. In this scenario, if the Redis server crashes, nothing of grave importance will have been lost.