com.ajjpj.asysmon.measure.scalar.AScalarMeasurer Maven / Gradle / Ivy
package com.ajjpj.asysmon.measure.scalar;
import com.ajjpj.asysmon.data.AScalarDataPoint;
import com.ajjpj.asysmon.util.AShutdownable;
import java.util.Map;
/**
* Scalar measurements are performed in two steps because some data (e.g. CPU load) must be calculated based on raw data
* from two distinct points in time. Things like 'load' is inherently an average over some period of time, after all.
*
* So first, prepareMeasurements() is called, allowing a measurer to store some data as a memento (e.g.
* the first number of jiffies together with the first timestamp).
*
* Then the caller waits some time, typically between a tenth of a second and a second, but potentially longer, to
* give measurers a solid time base to average data over.
*
* Finally, contributeMeasurements() is called, passing in the mementos of the preparation phase. This
* is when the actual data is collected. Non-averaging measurers can (and probably should) ignore
* prepareMeasurements() and the entire memento mechanism.
*/
public interface AScalarMeasurer extends AShutdownable {
void prepareMeasurements(Map mementos) throws Exception;
void contributeMeasurements(Map data, long timestamp, Map mementos) throws Exception;
}