Scale What Matters, Not Just CPU - Welcome Keda autoscaling

Scaling production workloads solely based on CPU and memory can be inadequate for systems like billing processes that rely on asynchronous jobs, where maintaining high CPU usage is preferable and scaling should align with task backlog. Qovery simplifies this by integrating KEDA into its platform, allowing for custom autoscaling based on specific metrics like queue length, thus reducing the need for manual adjustments and custom scripts. This approach significantly lowers the risk of delays during traffic spikes and ensures consistent adherence to service level agreements (SLAs), as demonstrated by a real-world scenario where a billing system successfully transitioned from CPU-based autoscaling to one driven by pending requests. DevOps teams benefit from decreased incident risks and less maintenance, while users are encouraged to trial this setup in their staging environments to see its efficacy with queue-driven workloads.