// the find
Stratio/sparta
Real Time Analytics and Data Pipelines based on Spark Streaming
Stratio Sparta was a declarative Spark Streaming pipeline builder — you described your workflow in JSON (inputs, parsers, aggregations, outputs) and it ran it on a Spark cluster without writing code. The project was officially discontinued in 2019 and the README says so on line one. Anyone finding this today is looking at a dead codebase.
- The plugin architecture was well thought out — inputs, outputs, and operators are genuinely isolated behind clean interfaces, making it straightforward to add a new sink without touching core logic.
- JSON-defined workflows with a job manager UI was a real usability win for data teams who didn't want to write Spark jobs by hand — the policy examples show it actually worked end-to-end.
- Broad connector coverage (Kafka, Flume, RabbitMQ, Cassandra, Elasticsearch, MongoDB, JDBC) was genuinely useful for 2016-era streaming pipelines.
- Stateful OLAP aggregations over streaming windows with configurable operators (mean, median, stddev, entity count) was non-trivial work and the cube abstraction held together reasonably well.
- Discontinued in 2019 — this is not a historical caveat, it is the entire story. The project is frozen on Spark Streaming, which is itself deprecated in favor of Structured Streaming.
- Built on Spark Streaming's DStream API, which Spark has been pushing users off since 2.3. The underlying compute model is gone, not just this wrapper around it.
- Depends on Akka, Spray (dead since 2015), and Mesos (largely abandoned in favor of Kubernetes) — the dependency graph alone makes this unrunnable on any modern stack without significant surgery.
- No migration path to the promised Sparta 2.0 — the email address in the README probably bounces, and there is no public successor repository to point to.