Originally from the User Slack
@Dylan_Piette: Hello to the scylla team,
I have a question for you
In our project we might have a use case with Scylla where we would be forced to do a lot of INSERT => READ => DELETE workflow with single rows
I was wondering how would you manage that ?
Wouldn’t it be preferable to create another boolean column named “Acknowleged” and have a batch job that will delete them later ?
So the new workflow would be INSERT => READ => UPDATE (Acknowleged = true) and later do a batch job consisting of DELETE WHERE Acknowleged = true
@dor: It’s hard to say, I’d start with a simplistic approach and just delete the row. Batching can primarily help if you have a window of time where the cluster is not busy and you can even call compaction later. It’s not a must
@Dylan_Piette: Ok thanks for your answer, I’ll go with that then
I was asking because I saw some posts online that says that cassandra like databases are not meant for these kind of use cases, and as we don’t want to deploy multiple databases system, I was wondering if there was a common strategies with scylla to deal with that
@dor: Deletes create tombstones (it basically similar to your ‘update’ and actual delete later
Most of the time we handle those tombstones well, so assume that’s the case
@Dylan_Piette: Ok great, thank you !