Skip to main content
The Streams builder lets you compose a Kafka Streams topology on a visual canvas — no Java project, no build pipeline. You connect sources, processors, and sinks, write the transformation logic as SpEL expressions, validate the whole graph, and deploy it. The deployed topology runs inside your Alginte instance, next to your cluster.

​
The building blocks

  • Sources read from topics — as a KStream (record stream), KTable (changelog view), or GlobalKTable (fully replicated table for joins).
  • Processors transform the flow. The palette covers the Kafka Streams operator set: map / mapValues / flatMap / flatMapValues, filter, selectKey, peek / foreach / print, merge, broadcast / split (branching), repartition, grouping (groupBy / groupByKey), stateful aggregation (aggregate, reduce, count, cogroup), windowing (time and session windows, suppress), and joins (stream–stream, stream–table, foreign-key) — with materialize for named state stores.
  • Sinks write the result to a topic.
Records flow along the edges you draw; where an operator needs logic — a predicate, a mapping, an aggregation — you write a short SpEL expression in a type-aware editor with completions and live validation.

​
The wizard

Creating a stream is a three-step wizard:
  1. Stream Properties — the application.id (the stream’s durable identity), serde defaults, and any Kafka Streams configuration you want to set explicitly.
  2. Stream Flow — the canvas. Add nodes, connect them, configure each node in its drawer. Validation runs as you build: nodes and steps carry error badges until the graph is complete and consistent.
  3. Submit Stream — review and submit. The topology is built server-side and starts running.
The canvas supports the editing conveniences you’d expect: copy/paste and undo/redo (Ctrl+C/V, Ctrl+Z/Y), multi-select with Shift+drag, and node duplication.

​
Where your work lives

  • Designs export and import freely — the topology (canvas layout included) round-trips as JSON. Exporting is your durable copy and your sharing mechanism; importing recreates the design ready to edit or deploy.
  • Deployed streams are ephemeral in the Community edition: they run normally, but a restarted Alginte instance comes up with no deployed streams — re-import and re-deploy to continue. The UI labels deployments accordingly at deploy time. Durable, auto-restored deployments are a Professional capability.
  • Processing state and progress always live on the Kafka cluster (changelog topics, committed offsets) — never inside Alginte. A redeployed topology resumes exactly where it left off; no data is lost by a restart.

​
Lifecycle

Deployed streams appear on the Streams page with their state. From the stream’s detail page you can start, stop, and pause/resume it, inspect the topology, and export the design.
A deployed topology runs inside the Alginte process. For scaling and multi-instance considerations, see Running multiple instances.