AWS with Docker and Kubernetes
We are running 3 Kubernetes clusters on AWS. AutoML/Stag/Prod. Hosting our images on AWS ECR. Every env has 6-7 micro-services consist of Rython, RabbitMQ, PostgreSQL and Tensorflow. We are mostly using it for handling Machine/ Deep Learning workload. We have a model server that is GPU accelerated and using Nvidia-Docker Autoscaling is implemented on every cluster using Horizontal Pod Scaling. Under the peak load, we cross over 60 EC2 instances. All the cluster are running on different Kubernetes version 1.9.8 and 1.10.x.