We have pinpointed a challenge within our real-time transcoding and rendering service where the autoscaling logic was not functioning optimally, leading to memory shortages. This occurred as the system attempted to scale up to manage a higher workload, but the newly created instances were over utilizing resources, causing them to fail. To address this, we have refined our autoscaling process. This adjustment ensures that new instances will use the available memory more efficiently, allowing the system to scale horizontally and maintain performance without overburdening the infrastructure