How Healthy Is My Infrastructure?
Slow page loads or excessive downtime can undermine even the best-designed, most effective, easiest-to-use website. While web analytics shows you what people are doing on your site; end user monitoring shows you whether they could do it—and how quickly they did it.
Availability and Performance
The most basic metrics for web health are availability (is it working?) and performance (how fast is it?), sometimes referred to collectively as performability. These can be measured on a broad, site-wide basis by running synthetic tests at regular intervals; or they can be measured for every visit to every page with real user monitoring (RUM).
In general, availability (the time a site is usable) is communicated as a percentage of tests that were able to retrieve the page correctly. Performance (how long the user had to wait to interact with the site) is measured in seconds to load a page for a particular segment of visitors.
What to watch: Availability from locations where visitors drive revenue; page load time for uncached and cached pages; end-to-end and host time at various traffic volumes; changes in performance and availability over time or across releases.
Service Level Agreement Compliance
If people pay to use your site, you have an implied contract that you’ll be available and usable. While this may not be a formal Service Level Agreement (SLA), you should have internal guidelines for how much delay is acceptable. Some SaaS providers, such as Salesforce.com ...