Whether it is a normal workday or Black Friday, service-oriented applications must be able to handle varying loads. This is the only way to ensure that users are provided with a good experience and that costs are kept to a minimum.

Kubernetes offers a way to vary the number of application instances running based on CPU or RAM utilization with the Horizontal Pod Autoscaler. However, modern applications often depend on a variety of components and should be able to respond to external events. These may include new messages in a queue or metrics in Azure Monitor.

As an application developer or operation manager, what do I need to consider to ensure that my application can respond to these events? How can I configure Kubernetes for “scale to 0” to run my application only when needed?

Using Azure Kubernetes Service and KEDA (Kubernetes Event-driven Autoscaling), this session will show with practical examples how to create and configure autoscalers to respond to external events and scale applications in Kubernetes accordingly.