From Sandbox to K8S: Deploying a Streamlit based object detection application using Minikube
“Before a software can be reusable, it first has to be usable.”
- Ralph Johnson
I am sure every dev-ops engineer out there will truly appreciate how much Docker technology has revolutionized the way people deploy applications by bringing in portability, flexibility, faster software delivery cycles and much more! Gone are the days where enterprises deploy their applications on virtual machines that consume gigabytes of memory and adding unnecessary overheads like setting up dependencies, services and configuration files for each application, thus bloating-up the operational cost. I am quite confident that we are already experiencing the next wave of deployment technology, and that is container orchestration.
In this series of articles, I have put together my experience in working with Kubernetes (K8S). It’s a popular open-source container orchestration technology that was initially developed by the software giant Google and later donated to Cloud Native Computing Foundation which currently maintains it. To be more specific, in this tutorial, we will explore how one can go about deploying a Streamlit based object detection model in Kubernetes. I assume that the readers have a basic working level knowledge on Dockers, if not, please refer to this series, one of the best Docker tutorial series I have read.
This series is organized into the following sections:
- Part 0: Prologue
- Part 1: What the heck is Kubernetes?
- Part 2: Streamlit based object detection application
- Part 3: Familiarizing the environment
- Part 4: The Crux
- Part 5: A Little Bit of Polishing
I have made the entire code that will be used as a part of this tutorial in my GitHub repository.