One useful feature of a stream processing engine is the possibility to get statistics about the performance of your running queries. Liebre gives you a basic set of statistics for your query. More specifically, it gives you statistics about:
If you want Liebre to keep these statistics, then write:
Query q = new Query(); q.activateStatistics(<STATS FOLDER>);
Please notice: The statistics are kept only for the streams, sources, operators and sinks you add to the query after invoking this method. This allows you to keep statistics only for some of the streams, sources, operators and sinks of your query.
All the statistics are produced as csv files that contain 2 columns. The first column contains the timestamp (UNIX epochs in seconds) while the second column contains the number of tuples added or taken from a stream or the per-tuple time (in nanoseconds) spent by a source, operator or sink.
For each stream with id X, Liebre will produce two files:
For each source, operator or sink with id X, Liebre will produce one file:
All the files will be produced in the folder you specify.