The same way you would use redis to cache your expensive dbms queries, you can use this as the 'redis' for S3 blobs.
We use it at my firm to cache frequently accessed ML models and training data. Getting these directly from S3 can take a few seconds per blob, and with the cache it takes that down to milliseconds.
