The following two-step procedure is a useful tool for analyzing/stress-testing network performance and can help pin-point Internet issues. It should be performed when the network is quiescent; that is, when there is little or no network activity. 

Checking your available Internet bandwidth

The computer running the tests described below should ideally be directly wired to the router or be connected as far upstream as possible over a wired (not wireless) connection.

This procedure does not take down or disrupt the Internet connection and only takes a few minutes to perform.

1. Speed Test. Please direct a web browser to speedtest.net and run five consecutive speed tests. Discard the first two results and report the last three results. Each result set will comprise three values for upload speed, download speed and ping-times. Please email IPFINITY support three sets of values (i.e. a total of 9 values).
2. Network analysis. Please contact IPFINITY Support for assistance.

Understanding your results

Normal speed test results will be consistent within a few percentage points of each other and be close to the ISP's specified bandwidth. 

If either (i) the speed results fluctuate significantly and do not trend towards the expected bandwidth numbers, or (ii) if the ping times are higher than a few tens of milliseconds, disconnect your internal network from the Internet, reset your router and/or other edge devices and repeat the tests from a computer directly wired to the most upstream device (usually the cable or DSL modem or router provided by the Internet Service Provider).

Should errors persist, please contact your ISP for a diagnosis of the issue.

Stress-testing your Internet with the Nightingale Installed

1. First obtain the quiescent speed-test numbers without the Nightingale connected and verify that they match the numbers provided to the IPFINITY provisioning team.
2. Now connect the Nightingale as specified in the technical documentation (available here).
3. initiate several, simultaneous downloads to achieve a minimum load of 90% of the available download bandwidth and simultaneously re-run several speed tests to measure the under-load speed-test numbers -- these are expected to be significantly lower than the quiescent numbers and may also show significant variability.
4. Place one or more phone calls while the network is under load.

If Nightingale is correctly installed, call quality should be clear even under 90% network load. Despite full-load speed-test numbers – signifying the remaining available bandwidth — being dramatically reduced in comparison to the quiescent numbers.

You may wish to perform an "A/B test" --- i.e. with and without the Nightingale "in the loop".