Sessions storage. Data modeling, sharding strategy

i’ve got couple of questions about my approach of using fdb.
The use case here is to store user’s sessions.
Keys looks like this:

Subspace(“sess”).Subspace(“ga”).Subspace(“random uuid 32bytes”) = around 200 bytes

There is also secondary key for tracking when it’s time to delete old sessions:

Subspace(“sess”).Subspace(“gc”).Subspace(“expire timestamp truncated to begginning of the day”).Subspace(“random uuid 32bytes”)

So it’s either insert or delete, no updates.

Is it gonna work in the long term? i mean for a very large number of keys?
Or will the perfomance degrade over time?
What about distribution of data, will it spread evenly across all storage servers?
Will any hot spots / hot shards emerge?

Hotspots could occur if the data is very small, i.e. small number of shards. Otherwise you should get great scalability as your data size grows. Assuming your workload isn’t dominated by a small number of keys.