Terms for Review

Node

A Node is a unit of storage in ScyllaDB. It is comprised of the ScyllaDB database server software running on a computer server — a physical machine — and all its subsystems (CPUs, memory, storage, network interfaces, and so on), or, in a virtualized environment, a subset of a server’s resources assigned to a container.

Cluster

A minimum Cluster typically consists of at least three nodes. Data is replicated across the Cluster, depending on the Replication Factor.

Table

A Table is how ScyllaDB stores data and can be thought of as a set of rows and columns.

Keyspace

A Keyspace is a collection of tables with attributes that define how data is replicated on nodes. It defines several options that apply to all the tables it contains, most prominently of which is the replication strategy used by the Keyspace. A Cluster can have multiple Keyspaces.

CQL

A query language for interacting with the ScyllaDB (or Cassandra) database.

CQL Shell

A command-line interface for interacting with ScyllaDB through the Cassandra Query Language (CQL)

Replication

The process of replicating data across Nodes in a Cluster.

Consistency Level

A configurable setting that dictates how many Cluster replicas must acknowledge read or write operations.

Tunable Consistency

The ability to choose the Consistency Level for each session or each query, trading consistency for latency and availability where that’s acceptable.

Replication Factor

The number of copies of each piece of data. With NetworkTopologyStrategy, it is set for each datacenter. A Replication Factor of 1 means that the data will only exist on a single Node in the Cluster, and this setup will not have any fault tolerance. The Replication Factor is set for each Keyspace. All replicas share equal priority; there are no primary or master replicas.

Tablet

A part of a table’s data, covering a range of tokens. ScyllaDB splits each table into tablets and balances their replicas across the nodes and shards of the cluster.

CAP Theorem

The CAP Theorem states that when a network partition splits a distributed database, it has to choose between Consistency (every read gets the latest write) and Availability (every request gets a response). In ScyllaDB, the Consistency Level decides that tradeoff for each query: a level that needs fewer replicas (such as ONE) keeps answering as long as one replica is reachable, possibly with stale data, while one that needs more (such as QUORUM or ALL) rejects requests when too few replicas are reachable.