Difficulty: Beginner
Estimated Time: 15 minutes

In this learning unit, you will:

  • Learn about tunable consistency and consistency levels
  • Use the CQL shell CONSISTENCY command to get and set consistency levels
  • Experiment with consistency levels ONE, TWO, THREE, LOCAL_ONE, QUORUM, LOCAL_QUORUM, and EACH_QUORUM
  • Understand how to choose appropriate consistency levels to meet your application requirements

This scenario is also available on our datastax.com/dev site, where you can find many more resources to help you succeed with Apache Cassandra™.

Did you know?

You can use Cassandra as a service in the cloud. Nothing to install, no credit card required. Sign up and launch your database with a few clicks at astra.datastax.com!

In this scenario, you learned about:

  • Tunable consistency and consistency levels
  • The CQL shell CONSISTENCY command
  • Choosing appropriate consistency levels to meet your application requirements

Tunable Consistency in Apache Cassandra™ and Consistency Levels

Step 1 of 9

Tunable consistency

Each keyspace in Cassandra has a data replication strategy, which prescribes how data should be replicated. There usually exist multiple copies of the same data stored on different nodes in a cluster. However, in the real world, nodes fail, go down, get replaced, go back up, become less responsive, and so forth. While some replicas may become unavailable, other replicas may just be temporarily out-of-sync. Given such conditions, will your application still be able to access data? Can it tolerate the risk of reading inconsistent or stale data?

The answer lies in tunable consistency. Tunable consistency is a mechanism that enables a per-operation tradeoff between consistency and availability. For example, for some writes, your application may prefer higher consistency to make sure that a large number of replica nodes get those writes successfully. For some reads, your application may prefer higher availability allowing the reads to be served by a small number of replicas making it possible to read even when many replicas are down. We will soon demonstrate how consistency can be tuned using consistency levels and discuss how to choose consistency levels to meet your use case requirements.

Our introduction of tunable consistency would not be complete without mentioning eventual consistency, the term that refers to replicas becoming consistent or synchronized over time. Since every column value has a write-time timestamp, it is straightforward to differentiate between up-to-date and stale data. Cassandra provides several mechanisms to synchronize replicas, including anti-entropy repair, read repair and hinted handoff.