1
00:00:01,041 --> 00:00:06,041
KOBE is controlled with a simple command-line interface.

2
00:00:07,148 --> 00:00:21,800
In a typical workflow, you begin by applying the benchmark definition,

3
00:00:22,336 --> 00:00:25,500
and after the source endpoints are initialized,

4
00:00:26,036 --> 00:00:32,036
you can apply the definition of the experiment to be executed.

5
00:00:34,094 --> 00:00:38,094
The benchmark and experiment definitions are simple YAML files.

6
00:00:39,000 --> 00:00:43,000
A benchmark comprises a set of datasets and a set of queries.

7
00:00:43,800 --> 00:00:46,800
An experiment configuration refers to a previously defined benchmark.

8
00:00:47,900 --> 00:00:51,900
The resulting metrics of the experiments are visualized in Kibana dashboards.

9
00:00:52,418 --> 00:00:56,418
In the first dashboard you can select an experiment

10
00:00:56,518 --> 00:01:00,378
and view several metrics for each query of this experiment,

11
00:01:00,800 --> 00:01:04,500
such as times of each phase of the query processing time,

12
00:01:04,250 --> 00:01:06,600
number of sources accessed and number of results.

13
00:01:09,925 --> 00:01:14,925
The second dashboard is used for comparing experiment executions.

14
00:01:16,225 --> 00:01:21,225
You can select all the experiments of interest,  

15
00:01:22,590 --> 00:01:26,590
and compare the query processing times and number of returned results.

16
00:01:33,689 --> 00:01:38,689
Finally, in each experiment execution, we can have several runs of each query.

17
00:01:38,789 --> 00:01:43,789
Thus, in the third dashboard you can select a specific query,

18
00:01:43,889 --> 00:01:48,889
and several experiment executions that contain this query.

19
00:01:48,989 --> 00:01:53,989
You can compare all runs of this query for each experiment execution.

20
00:01:57,000 --> 00:01:59,000
Thank you for your attention.