High Availability Deployments at Twilio Flex - a Case Study

Abstract

It is important for Twilio Flex to minimize downtimes and the blast radius for breaking deployments to ensure that the Service Level Agreements (SLAs) are not violated, and a minimum number of customers are affected for a minimum amount of time. This paper has set a goal of implementing a Proof of Concept (PoC) of high availability deployments in Twilio Flex with this motivation. In order to achieve this goal, a method was set out to select one of the Twilio Flex services for the PoC. Also, a suitable tool for high availability deployments was selected. Then, the PoC was designed, implemented, tested, and the results were evaluated. The final evaluation shows that the PoC did not significantly affect the selected service’s reliability, robustness, and speed. It was concluded that utilizing feature flags for gradual rollouts can bring benefits to high availability of Twilio Flex.