Why does a Cassandra cluster need a synchronized clock between nodes?

Question

The Cassandra DataStax introductory course says that all the clocks of the Cassandra cluster nodes must be synchronized to prevent READ requests from the "old" data.

If one or several nodes are omitted, they cannot receive updates, but as soon as they return again they will be updated and there are no problems ...

So, why does a Cassandra cluster need a synchronized clock between nodes?

Answer 1

, , , , Cassandra "Last Write Wins" , . .

, "" ( ) () cassandra, , . .

, cassandra , , , , ( , , , ).

, . , node, , . , , , , .

, :

, 2 node A B. , , A t10, B - t5.

• DELETE C FROM tbl WHERE key=5. node t10.
• , UPDATE tbl SET C='data' where key=5. node B t6.
• SELECT C from tbl where key=5. DELETE 1 (t10 > t6), .

, datastax , , C * . datastax java-driver 3.0 ( " " ). , , , , cassandra, .