Perception is reality on the web. How users perceive the site to be responding becomes the fact. This has been proven when looking at healthcare.gov. We are getting a lot of questions regarding the variances in performance across the different states in the US. If you look at the following performance hotspot map you can see that certain states are performing really well (Green means faster than 4 seconds) while others are performing really bad (Red means slower than 8 seconds).
Huge difference in page load time can be explained by latency and bandwidth
Comparing two states
Comparing the calculated Key Performance Indicators for these two test runs shows the difference in “hard facts”. Bad latency and low bandwidth results in very high Connect Times to establish a physical connection. Before connecting it also impacts DNS lookup times. Downloading the actual content (Transfer Time) as well as calculated server-time is impacted as well by low bandwidth. Why is Server-Time impacted? Because the server time is also impacted by latency and bandwidth. The initial response takes longer on high latency and slow bandwidth which impacts the calculated server-side time:
DNS, Connect and Transfer are the Performance Indicators that have the biggest impact on bad page load time
How to fix this?
Obviously nobody can just magically upgrade network bandwidth or lower latency of the actual end users. But – by addressing all the “homework steps” we highlighted in our previous blog posts (Diagnose Obamacare Website, Practical Testing Tips and Live Healtcare.gov Performance Report). Making these pages “slimmer” and with that reducing the payload as well as the number of roundtrips will lessen the impact of bad latency and low bandwidth.