Unified Workload Management Through Kubernetes-Native APIs

Abstract

Enterprise IT organizations must bridge the architectural gaps between their legacy virtual machine workloads and their cloud-native container-based applications. Virtual machine workloads depend on isolation for reasons such as operating system dependencies, regulatory compliance, or other characteristics such as performance, and microservice architectures are typically deployed on a container orchestration layer for agility and scalability. This article presents unified workload management patterns for declarative, Kubernetes-style deployment and lifecycle management of virtual machines (VMs) along with container workloads on a common control plane. It also presents an architecture that can support key enterprise workload patterns such as workload portability, multi-tenancy governance, developer self-service, and operating model portability across heterogeneous VM and container compute infrastructures. Drawing on experience of running production workloads on both VMs and containers on the same infrastructure, this article shows how Kubernetes native APIs can be extended to provision, configure, and manage VMs in the same declarative way as containers, breaking down infrastructure silos, preserving enterprise security and compliance, and allowing organizations to run faster without sacrificing operational stability. This helps enterprises achieve their digital transformation goals while also protecting existing virtualization investments.