Skip to content

measure deltas between timing events #14

@sbeam

Description

@sbeam

This issue is more to get your thoughts on this as a feature.

We are seeing very data coming into our InfluxDB collector that is so noisy as to make it near impossible to get actionable metrics. For instance, domInteractive on a certain page has a mean of 8329 with a standard deviation of 13898. (sample size=62000+). Which is due to some very large long-tail response times - but even throwing out everything above the 90th percentile, the stddev is still at 6188.

They way these are collected as offsets from navigationStart is good, but it seems to me the variation in each of these measurements from that point is cumulative. In other words, domInteractive is more variable because it includes all the potential variations in caching, DNS lookups, response times, etc.

It might be helpful to collect offsets between each timing event and it's predecessor as well. For instance, performance.timing.domInteractive - performance.timing.domLoading would be an interesting handle on blocking CSS or JS issues.

It may be something that could be left to the user as a config option.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions