In this chapter, you learned how to set up a multinode cluster by changing various configuration options. We also took a practical look at how consistency levels work and how Cassandra provides a balance between consistency and availability. We tried out various consistency levels including QUORUM and ANY for writes. We then took a quick look at the architectural aspects of Cassandra.
We looked at the write path, and how data was written to both memory and disk. Data was persisted to commitlog on disk to avoid data loss in case of restarts. Data is flushed to immutable SSTables when the memtables are filled up. The read path utilizes several data structures, both in memory and on disk, to optimize reads. We could enable row and key caching to avoid disk seeks. In case a partition was not found in cache, we would have to hit bloom filters and partition indexes to figure out the location of a partition within an SSTable. We also looked at the data repair mechanism provided by Cassandra...