Flume events can be passed across multiple agents over the network. Flume uses a Transactional approach to guarantee the reliable delivery of events. The Flume source(s) and sink(s) encapsulate, in a transaction, the storage and retrieval, respectively, of the events. This ensures that the set of events are reliably passed from point to point in the flow. In the case of a multi-hop flow, the sink from the previous hop and the source from the next hop both have their transactions running to ensure that the data is safely stored in the channel of the next hop.

In addition to providing Transactional capability, the flume channels can be executed in persistence mode. This means that the data will be stored on disk by the channel so that an abrupt shutdown of the Agent does not cause a loss of data. Further reliability is built into Flume by providing features such as built-in Load-Balancing support and built-in Failover support.