Export of observed values

7. Export of observed values

JISD has a function to export the observed values in real-time. Currently, the following tools are supported for exporting.

Elasticsearch
Prometheus

We recommend Elasticsearch as the export destination for JISD. If you use Elasticsearch as the export destination, JISD will cache the observation values and send them at regular intervals, even if the observation values are updated frequently. JISD caches the observed values and sends the data at regular intervals, so the values are not lost.

Users can visualize the observed values using Grafana or Kibana from the above tools. At this time, please pay attention to the display interval of each visualization tool. JISD will issue a warning if it detects too short an interval, but the value observation will continue as usual.

In the following sections, we explain how to visualize the observed values using Elasticsearch and Grafana.

In the following, you will need the Elasticsearch and Grafana environments. Demonstrations of these environments are defined in docker-compose.yaml in JISDLab/JISDVis. Please follow the instructions in JISDLab/JISDVis/README.md.

7.1 Exporting to Elasticsearch

First, specify the program to be observed jisd.demo.BinarySearch as the observation target.

var dbg = new Debugger("jisd.demo.BinarySearch", "-cp ../sample");

Next, we create an instance of ElasticsearchExporter, a class that exports data to Elasticsearch.

// ElasticsearchExporter(host, port, name, timeLocale = "00:00")
var esExporter = new ElasticsearchExporter("http://elastic01", 9200, "sample");

In addition to the host name (host) and port number (port), it is necessary to give a name to the observation data set. The name will be used as the id name of Elasticsearch to store the observation data.

To link this esExporter to the observation target, do the following.

dbg.setExporter(esExporter);

We are now ready to export the observations. Let’s try to export the movement of variables left and right in binarySearch() of jisd.demo.BinarySearch

// Set the observation point
dbg.watch(19, new String[]{"left", "right"});
dbg.watch(25, new String[]{"left", "right"});
esExporter.run(10); // start esExporter (sleep for 10ms for each observation point)
dbg.run(3000); // start Debugger (start the program to be observed)
esExporter.stop(); // stop esExporter

If you start esExporter after setting an observation point, it will continue to send data to Elasticsearch every second until it stops. esExporter.run() can be overloaded by sleeping for ms specified in the argument of esExporter.run(). This is used when the interval between value observations is too short to be displayed properly in the visualization tool. This is used when the interval between value observations is too short to be displayed well by the visualization tool. If no argument is specified, the value observation can be performed without loading.

7.2 Visualization with Grafana

7.2.1 Adding a Data Source

HTTP
- URL
  - Enter the URL of Elasticsearch (in the demo, http://elastic01:9200/).
Configure Elasticsearch details
- Index name
  - Enter the string specified in the argument name when creating an ElasticsearchExporter instance.
- Time field name
  - Enter @timestamp.
- Version
  - Select 7.0+.

Finally, click on Save&Test to save the configuration.

7.2.2 Adding a dashboard

In the dashboard edit screen, enter the data source you just created in the New Dashboard Data Source field.
in the Query field, type name="<variable name>" to specify the variable name you want to display (for the demo, type name="left" and name="right" in separate queries).
select Average in the right column of Metric and value in the Select Field field
select Last 5 minutes in the time ranges field on the top right of the dashboard edit screen

7.2.3 Value Observation and Visualization

Again, follow 7.1 from above to observe the values and export the data. After that, Grafana will reflect the observed values.

JISDLab

Installation