Datastax has completely dropped the ball with Cassandra. Years of unusable builds because of decisions they couldn't even begin to satisfy or test properly.

Cassandra is a great database design but the implementation has never really matured and become a polished product. The datastax distribution was a nice package but suffered from complexity and poor performance. The political battles also hurt by making even more variations of the database which confused users.

Cassandra is even starting to use some of the architectural ideas from Scylla but it's stuck in a messy state so far. Scylla has the opportunity to cleanup and make forward progress with a solid technical foundation.