Topics define a data stream. You can think of a topic as the streaming equivalent of a table in a database. Topics can be parallelized into any number of data slices called partitions.

In addition to logically separating records, topics may be individually configured with parameters to tune the performance and semantics of record delivery.

Topics also have a replication factor that defines durability, the number of copies for each data slice.

Replicas have a leader and one or more followers and distributed across all available SPUs according to the replica assignment algorithm.

For example, when provisioning a topic with 2 partitions and 3 replicas:

$ fluvio topic create topic-a --partitions 2 --replication 3

Leaders maintain the primary data set and followers store a copy of the data. Leader and follower replications are assigned to independent SPUs:

Leader, Followers

  • topic-a/0 is the first partition
    • leader on SPU-1
    • followers on SPU-2 and SPU-3
  • topic-a/1 is the second partition
    • leader on SPU-2
    • followers on SPU-1 and SPU-3