It would be better to update to add new or changed data. There is no need to rewrite old data that does not change, and it would be inefficient to force Cassandra to rewrite it.
When you insert or update, Cassandra saves the change timestamp for each column. When you read, Cassandra collects all records for this key from memory, from disk and other replicas depending on the consistency setting. It will then combine the column data so that the newest value is used for each column.
When data is compressed on disk, if there are separate updates for different columns of a row, they will be merged into one row in the compressed data.
You do not have to worry about creating tombstones by performing an update if you are not using the update to set the TTL (Time To Live) value. In your application, it sounds like you never delete data, so you will never have tombstones.
Jim meyer
source share