Run Gobblin as a standalone application on a single box or JVM, or simply run in embedded mode with your application.
Enterprise: Big Data
Run Gobblin as a map-reduce application on multiple Hadoop versions. You can also leverage workflow schedulers like Azkaban to launch MR jobs.
Run Gobblin as a standalone cluster with primary and secondary nodes on Yarn or Kubernetes. This mode also supports HA.
Enterprise: Cloud Native
Run Gobblin as an elastic cluster on public clouds like AWS (Azure and GCP support coming soon). This mode also supports HA.