Scheduled Tests

Device Agents support the scheduled execution of tests. A typical test schedule involves executing some tests once per hour, and other heavier tests (such as the speed tests) once every few hours. Alternatively, clients may wish to focus their measurements during the peak-hour periods to gain the most representative view of user performance. See Test Triggers for examples of test schedule use cases.

If CPU load is high or cross-traffic is detected when a scheduled test is due to be executed then the device will enter a pause-and-retry loop, checking threshold levels every five seconds for up to 30 seconds. If the levels of cross-traffic or CPU load are still above thresholds after 30 seconds has elapsed, then the scheduled test is skipped and this event is recorded, along with which threshold was breached and the breach values.

While the same test schedule can be assigned to thousands or even millions of devices, that does not mean that they all run measurements at the same instant. To do so would risk overwhelming the network or the test servers. The test schedule includes a window within which tests should take place. Our platform then randomizes measurements within that window to spread out the measurements before generating a test schedule for each agent. This means that each agent receives a customized test schedule that ensures that its measurement traffic does not coincide with too many others.