Vibe Data Stream for Machine Data (VDS) does not stage data at any point between the data source and the target. To transport data, VDS uses the publish/subscribe model. VDS generates a topic name for a data flow. Source services read data in blocks and publish messages on a topic. Target services subscribe to that topic, and they receive those messages and write them to the target data store.
You can configure source services to use streaming or load balancing messaging mode to publish data.