Here at StatusCake we’ve always pushed the idea that with website monitoring every second counts, be it reputation, revenue or trust all can be lost. It’s with this in mind that we have made some changes to our test systems today.
Previously our system would do confirmation tests after a test has failed and this means if your site is stuck loading (for example it’s under heavy load) it can take up to your configured timeout rate for each confirmation. If you have set to confirm with 4 servers and each one days thirty seconds then you won’t get alerted for over 2 minutes!
From today when a test is taking longer than its average load time the confirmation servers will start work right away all at once, if the original test completes without a problem just a bit slower than normal then the confirmation data will be dropped, however if the original signs of a problem end up becoming a downtime trigger all confirmation servers will have already started (if not completed) the work load they had been set.
This means we’re shaving of seconds to when you first know about an issue affecting your website all the while without introducing possibilities of false downtime.
Along with these changes we’ve made some design decision changes. Previously your uptime would be determined by any downtime beyond that of the designated trigger rate you had set, so if you had downtime of 4 minutes with a 5 minute trigger rate then your uptime would remain at 100% – now uptime is taken from the first validated confirmation. This change was made because we believe it better reflects the real world data.
Both these changes are part of a continuous ongoing improvement program we have here at StatusCake. Stay tuned for more improvements (and new features!)