Partitioning, consistency levels, query-first modelling and compaction.
In Cassandra, running DELETE on a row does not remove the data from
disk straight away. What does it write instead, and why does Cassandra
delete data this way?
What is compaction in Cassandra, and what problem does it solve? Give the main things it does to the data on disk.
Your team stores one-time login codes with INSERT ... USING TTL 600.
What does TTL do, and what happens to the row in Cassandra once the ten
minutes are up?
What is gc_grace_seconds on a Cassandra table, and what is it for?
A teammate's code always inserts every column of a users table and
binds null for fields the user left blank, such as middle_name.
What does Cassandra actually store for those nulls, and what would you
change?
You ran a script that deleted about half the rows in a large Cassandra table, expecting disk usage to drop. An hour later the disk is slightly fuller than before. Why, and when will the space actually come back?