Cassandra Practice Test

In which scenario would a new partition on disk be larger than its input partitions after compaction?

1. When the input partitions consist mostly of DELETE operations

2. When the input partitions mostly contain UPDATE operations

3. When the input partition segments are made up of most INSERT operations

4. When the input partitions are of equal size

The correct answer is: When the input partition segments are made up of most INSERT operations

The scenario where a new partition on disk is larger than its input partitions after compaction occurs primarily when the input partitions are composed mostly of INSERT operations. In Cassandra, an INSERT operation creates new data in the table, which increases the size of the storage even if it replaces existing data. When compaction runs, it merges the input partitions into a new, consolidated partition. If the input partitions include numerous INSERT operations, the amount of new data being written can exceed the size of the original data, particularly if the INSERTs are adding additional fields or new rows that were not present in the previous partitions. This means that even after removing tombstones (from deletions) or duplicates (from updates), the overall size can still increase. In contrast, scenarios where there are many DELETE operations lead to tombstones, which generally do not increase the size but instead mark previous entries for removal. Similarly, when the input partitions primarily consist of UPDATE operations, the added data typically replaces older data, which may not significantly change the size. Finally, if the input partitions are of equal size, the outcome after compaction won't increase the size as significantly, since it balances out the data without introducing much new information.