Hi Otis,
Suppose you hav written an application that needs to send data via Flume - you’d use the client SDK’s APIs to write data to Flume. The APIs are wrappers around Flume’s RPC protocol - so it sends the data to Flume in a much more efficient way than HTTP etc. The Client SDK provides an API to send data to Flume, not receive data from Flume. For that it is better to just write a sink that talks the external store’s native protocol. Thanks, Hari On Fri, Oct 31, 2014 at 10:15 AM, Otis Gospodnetic <[email protected]> wrote: > Hi, > I recently noticed somebody (Ashish, I believe) mentioning Flume SDK. > Never heard of it before. Googled, but really only found this: > https://flume.apache.org/FlumeDeveloperGuide.html#client-sdk > Is there any other info/documentation available about this anywhere? > What's the main use case for this? If there are no Flume sinks or sources > for one's data stores or end-points, then one can use Flume's data > transport mechanism by sending data to Flume via RPC... is that the main > idea? > Thanks, > Otis > -- > Monitoring * Alerting * Anomaly Detection * Centralized Log Management > Solr & Elasticsearch Support * http://sematext.com/
