Using Apache Cassandra as a backend for a queue or queue-like structure is never going to end well. We have discussed at length that, due to Cassandra's log-based storage engine, inserts, updates, and deletes are all treated as writes. Well, what does a queue do? It does the following:
- Data gets written to a queue.
- Data gets updated while it's in the queue (example: status). Sometimes several times.
- When the data is no longer required, it gets deleted.
Given what we have covered about how Cassandra handles things, such as in-place updates (obsoleted data) and deletes (tombstones), it should be obvious that this is not a good idea. Remember, a data model built to accommodate a small amount of in-place updates or deletes ...
