October 2018
Beginner to intermediate
348 pages
10h
English
As a Cassandra DBA, this is a common problem for which you will have to monitor your logs. All developers have been taught that batching-up tens of thousands of writes together helps performance; in the RDBMS world, it does. Inevitably, a developer will try this on one of your clusters, and could possibly cause the node to crash.
Remember, BATCH in Apache Cassandra is a misnomer. It should probably have been named ATOMIC, because it ensures that all writes within a batch are applied atomically. The idea is that four or five query tables can be kept in sync using BATCH, so that a failure of one write will cause all of the others to be rolled back. It really was not designed to help with performance; in fact, you take ...