I need to place some data into redis so that each element has a sliding expiration (say 60 seconds each).
I also need to be able to get a list of all elements that are currently in Redis.
Initially i just placed all elements in one-by-one, but unfortunately when it came to loading them, i realised that
SCAN() have a huge overhead, both in terms of clustering (e.g. keys only works on a server in the node) and in terms of real time performance on databases of a few billion records.
The next thing i looked at are sets, the problem being that i have to iterate the sets myself, and expire items myself, too.
What's the best practice for something like this?