When investigating a website slowdown, you first want to determine whether the slowdown is originating within the Section platform or at the origin server. The following guide will help you create a visualization of website performance data.
The Kibana view can be opened up in a new tab by clicking the “Open in a new window” button. This is useful if you wish to share the visualization with a colleague.
You will now configure the X-Axis to display the data over a period of time.
You will now configure the Y-Axis to display the average time taken in milliseconds for the edge logs.
You are now viewing the average time_taken_ms value over your selected time frame for all logs. Next steps will be to split the lines by log type to pinpoint the cause of the slowdown.
Here is an example of the finished product with clear increases in LastProxy access log times. This is a strong indicator for a slowdown at the origin application:
If you see a pattern of clear increases in LastProxy time_taken_ms values that resembles the above screenshot, the slowdown is likely being caused by the origin server and not Section. From here, look into origin server logs and an origin server monitoring solution such as New Relic if available.
If the above investigation reveals increased time_taken_ms values for Edge, Varnish Cache, or another module in your stack but a stable LastProxy value, this could indicate an degradation of service within the Section platform. In this case email firstname.lastname@example.org to immediately receive support from an engineer.