VoltSP Pipeline Architecture

Sources ingest streams from Kafka, Pulsar, MQTT, HTTP, S3 (with SQS event notifications), files, Beats, network sockets, generators, and stdin — all fan into a single Source operator.
VoltSP runtime executes a fluent chain of processors (Java, JavaScript, Python, ONNX inference, Parquet, VoltDB cache) with at-least-once semantics, plus emitters & windowing (tumbling time/count, VoltDB bulk-procedure, VoltDB procedure).
VoltDB state store sits on top of the runtime — every processor can read and write deterministic state for each event on a 3-node ACID cluster (in-memory, serializable).
Sinks deliver to 14+ downstream systems. VoltDB-bulk-insert and VoltDB-procedure are highlighted as the first-class transactional egress paths.
Deployment is cloud-native — one declarative pipeline spec (YAML or Java DSL) runs on Kubernetes via Helm or as a plain JVM service on bare metal.