Jepsen: Scylla 4.2-rc3
p) triples called cells, but conceptually, a Scylla row is a map of column names to values. Those values may be primitives (e.g. strings, integers, booleans, dates), counters, or collections, such as maps, lists, or sets. Collections are internally stored using multiple cells. Rows are grouped into partitions by a partition key, and those partitions are assigned via a Dynamo-style hash ring to a subset of nodes in the cluster. Each partition is replicated across multiple nodes for redundancy. Like Cassandra, Scylla generally allows clients to write to any node at any time—even when node...
Source: jepsen.io