From today's Q and A session, 08/28/2020, I understood that SSTables can be removed even with the cluster running. Did I get it right? That is really cool for time series data.
So, there is no metadata tracking SSTables in Cassandra?
This question was
closed
by
Erick Ramirez for the following reason: The question is answered, right answer was accepted
Yes, Cassandra keeps track of the SSTables it manages. I feel that you've taken Patrick McFadin's comments out of context so allow me to explain.
Cassandra keeps track of all SSTables it owns. You will see on startup in the debug.log that C* opens all the files so it knows what data is on disk including the partition index, partition summaries, etc.
For the purposes of time series data, we recommend you use TimeWindowCompactionStrategy (TWCS) with a TTL on the data. When all the data in an SSTable (time window) is fully-expired, all TWCS does is drop (delete) it from the filesystem. Cheers!
5 People are following this question.
