Workload types

A workload is something Streamtime runs on a fleet for you. There are three kinds, and one fleet can run any mix of them:

Kind	What it runs	Sized in	Platforms
Streaming	Kafka-compatible clusters your applications produce to and consume from, plus Kafka Connect, Kafka Replicator and virtual Kafka clusters	Kafka Units	Apache Kafka, Confluent (CFK), WarpStream, Redpanda (preview), AutoMQ (experimental), Confluent Platform on VMs; LeafCuttr and Tansu (experimental, not provisioned yet)
Stream Processing	Apache Flink: Job Workspaces you submit jobs to, single-job clusters, and Flink projects that hold your code	Flink Units	Apache Flink, Confluent Flink
Analytics	ClickHouse, the columnar database fed from your streams	ClickHouse units	ClickHouse

Every workload has the same lifecycle in every interface (console, API, CLI, MCP and, where a resource exists, Terraform): create it on a fleet, check its status, change or scale it, upgrade its Version, monitor it and delete it. Each workload’s page walks through these tasks.

A fleet only runs the kinds of workload it was set up for, with the capacity it was given for each; see Capacity units.

See Workloads for every platform’s page.