Skip to main content

Chasing Waterfalls - Resource Timing Data in the Wild

Mike McCall (Akamai Technologies Inc), Mark Holland (Akamai Technologies Inc)
Performance
Mission City Ballroom B1
Average rating: ****.
(4.17, 12 ratings)

In the world of web performance, Akamai is uniquely positioned. Between 15-30% of the world’s web traffic flows through our network at any given time, which gives us unique insight into how ISPs, devices, and browsers ‘see’ the web. With the recent implementation of the W3C Resource Timing specification in major web browsers, that insight has become even clearer.

For anyone who has used a synthetic monitoring tool like WebPageTest, they know the value of a web page waterfall chart, which illustrates the page load process. Now that Resource Timing is supported by the most recent versions of Chrome and IE browsers, we are able to understand what a waterfall looks like at RUM scale.

This presentation will cover the following topics:

  • Resource Timing
    • What is Resource Timing?
    • What is a waterfall?
  • RUM Resource Timing at Akamai
    • How did we get this data?
    • How did we analyze it?
  • Deep dive into Resource Timing data from a number of different sites and market segments
    • Resource Timing waterfalls at various percentiles
      • Do page loads around the median look the same? What about at the 90th percentile?
    • Identifying “slowness”
      • What are the reasons an end user falls into these percentiles? Can we attribute it to a specific cause or design pattern (or two, or three!) by analyzing the waterfall?

Attendees will be able to take some important concepts away from this talk, including what real-world performance patterns emerge when we look at a web page load holistically, and how they can leverage Resource Timing data within their own RUM implementations.

Photo of Mike McCall

Mike McCall

Akamai Technologies Inc

Product Architect in the Web Experience division at Akamai. Previously worked as a Senior Performance Engineer working on Real User Monitoring and data analysis.

Photo of Mark Holland

Mark Holland

Akamai Technologies Inc

Lead a team of Performance Engineers at Akamai, focusing on new product lines for whole-site delivery. Prior to that, spent many years in the data storage industry. Ph.D. from Carnegie Mellon University in 2004.