Cloud is the de-facto standard for data driven applications. However, cloud provider services tend to be slow, expensive and come with potential lock-ins. With emerging open-source cloud-native orchestration tools like Kubernetes, open source data stores and serverless platforms you can build your own multi tenancy cloud data service, save costs and gain performance and portability. The session will describe how you can build an equivalent set of AWS data services with cloud agnostic tools and achieve an Amazon like experience. Present a variety of analytics tools and explain how to use Kubernetes, as well as how to handle . A lesson learned through the suffering of others.