Frequently asked questions (FAQ)
Kaskada® is a unified event processing engine that provides the power of stateful stream processing in a high-level, declarative query language designed specifically for reasoning about events in bulk and in real time. Kaskada bridges the gap between batch and real-time event processing. Kaskada can match model context in training and production for better performing models—without costly code rewrites.
DataStax Luna ML gives you access to Kaskada expertise at DataStax as you run your own open-source deployments, but with the peace of mind that comes from having direct access to experts.
You don’t need Luna ML to use open-source Kaskada; however, Luna ML offers support for your Kaskada deployment from the Kaskada experts at DataStax.
You can learn more about subscribing to Luna ML on the Luna ML product page.
Luna ML supports all unmodified Kaskada open-source components.
Luna ML can help you deploy Kaskada with Cassandra and Pulsar.
Depending on your requirements, you can use Cassandra and/or Pulsar as sources and sinks with Kaskada.
Cassandra can be an event store (source), feature store (sink), and prediction store (sink). Cassandra offers balanced write and query scalability, low latency, and reliability while supporting large data volumes required by machine learning at scale.
Pulsar can stream new data into Kaskada as a source and Kaskada can write query results by creating a materialization and writing to Pulsar as a sink in real-time.
Luna ML is a support subscription for open-source Kaskada. Documentation for Kaskada can be found at kaskada.io.