KEDA, Kubernetes Event-Driven Autoscaling, is an open-source project enabling automatic scaling of resources in Kubernetes based on event-driven triggers, contrasting with the classic HPA limited to CPU and memory usage. It allows Kubernetes workloads to dynamically scale based on various events such as custom metrics from third party software (Prometheus, RabbitMQ...), making it well-suited for optimising resource utilisation.
During this session, we will dive into KEDA, what it is and how it works. Then, we will cover our use case, using KEDA to perform event driven autoscaling down of resources. First, we will review our decision to use KEDA and the alternate options. Then, we will explore metrics exposure using Prometheus, and how to read them with KEDA. We will then move on to the Keda setup and the main challenges we faced. Finally, we will evaluate the anticipated cost savings and compare it to the actual results.
All the steps will be illustrated by a live demo!
Solene Butruille
Solene Butruille is a software engineer at BlackRock working on Aladdin, a software that unifies the investment process. Solene has been working on multiple areas of the software including cybersecurity, alpha generation and cloud native applications.
She became passionate about new technologies during her studies in France and India. She is always eager to learn more and share knowledge with others! She did a few talks for BlackRock developer's community about various topics and was a speaker at Devoxx UK in 2023.