Much like the traffic light at an intersection inside of a city, a wait point in software is a point at which work queues due to limited access to a resource – often a third-party system.
As with a traffic light outage, if order processing is slow or offline customers will be left waiting. This effect ripples through other streets, creating traffic jams throughout the city.
Like intersections and traffic lights in a city, there are myriad wait points in a typical ecommerce site. The average hybris store has about forty of these wait points. A slow down in one area often leads to bizarre, ripple effects in other areas of the site. For example, slower verification of mailing addresses in turn slows down the performance of the homepage.
The Peak Ready Performance Approach
The Peak Ready Performance Approach begins by identifying all of these wait points, as many are unique to a particular online store. We then test each wait point in healthy, slow and failure modes using two Tacit Knowledge-built open-source technologies.
First is the Simulator, which mimics the behaviour of any third party system complete with real-world failure scenarios
Second is Slow Light, which acts like molasses, slowing down the speed of each response from the Simulator
Two load tests: slow and stop, are run for each wait point, usually about eighty in total.
We carefully monitor the site’s behavior, and modify the supporting software so that it becomes resilient under these conditions.
For example, should a payment gateway for credit cards experience an outage, the online store would either accept alternate payment methods only or gracefully queue orders for later processing.
Traffic would be re-routed around the obstacle until connectivity was restored.
For sites already developed or under active development, Peak Ready Performance can be used to identify a list of all wait points, catalog their behavior under exceptional conditions and describe recommendations for remediation
Tacit Knowledge also provides professional services to implement the recommendations and recertify the online store.
Real World Example: Beaverbrooks
For a real-world example of the results of this approach, one can refer to Beaverbrooks – a luxury brand eCommerce site that relies heavily on rich, photographic imagery.
We wanted to assure a level of quality commensurate with this iconic, luxury brand. Beaverbrooks has maintained an uptime of 99.996%, and an average of one-second page response times despite traffic spikes in excess of four times normal.
This is an example of why Tacit Knowledge applies the Peak Ready Performance approach to every ecommerce store we build.