Sources-Avro (data serialization with a schema), Thrift (abstraction written by Facebook to facilitate communications between different programming languages), Exec (shell commands), JMS (messaging), Twitter, Kafka (LinkedIn’s streaming product), NetCat, Syslog, HTTP, custom (you can write your own, plus there are plenty of third party ones.) Some of the sinks, sources, and memory channels that Flume supports include: And a memory channel stores events in memory, with the proviso that those get lost if the agent dies. Flume source and sinksĪ source is input data.
Instead you start agents, one for every data source that you want to read and save. There is no classpath or anything like that to set.
Use the right-hand menu to navigate.) Installation (This article is part of our Hadoop Guide.
0 kommentar(er)
