Improve pipelining during data-read, data-parse, parsed-write.
Currently metrics-lib isn't using system resources effectively. The steps data-read+data-parse block waiting for the parsed result. The result is then handed over to a queue for parsed-write. This happens in a loop until all data is read. These three steps form a single-core bound execution.
This ticket documents the effort to improve the way these steps are performed.
Trac:
Username: leeroy