It's funny, but that was my first thought as well.
Back on topic though, In practice it's seeming that more and more that these are very hard problems, and depending on your use/needs you may be better served by a more traditional replica-set data store with firm acid compliance over distributed systems...
I like distributed systems... Cassandra, RethinkDB and even MongoDB, Azure Storage Tables, ElasticSearch and Etcd are all pretty cool with their advantages/disadvantages... all of that said, take your real needs in terms of scope, scale and time into consideration before investing heavily into an always-available system and if you really need it.
