IBM have released the industry's first bare metal Kubernetes managed service, available via their IBM Cloud Container Service. This service will enable developers and data science teams to deploy containers to a fully-managed Kubernetes cluster running directly on bare metal cloud infrastructure.
Developers have faced challenges in the past due to the fact that installing and operating Kubernetes required a significant amount of configuration and management. However, running Kubernetes as a managed service requires less developer expertise and time commitment in order to operate a healthy cluster, due to features like automated provisioning and updating, and the intelligent scaling of the underlying infrastructure.
With Kubernetes running on bare metal, developers using the IBM managed service will get the benefit of having dedicated infrastructure for running their containers. Not sharing the underlying compute fabric provides increased hardware isolation, improved security and better performance. IBM’s CTO for Watson and Cloud, Bryson Koehler said in a TechCrunch article on the launch of Kubernetes on bare metal:
This now also opens up the ability to attach GPUs to these machines, which in turn enables the kind of machine learning and high-performance computing workloads that many enterprises are now starting to experiment with. If you look at the types of workloads that enterprises are moving to the cloud, bare metal is a huge benefactor in terms of isolation and flexibility.
IBM has supported Kubernetes since its launch, and has recently contributed to efforts to enhance the security and scalability of the platform. An example of this includes the work IBM have undertaken with Google and Lyft in order to build Istio, a service mesh control plane that aims to provide "an open platform to connect, manage, and secure microservices". A further example of their contribution includes bringing Kubernetes to bare metal. Jason McGee, VP of the IBM Cloud Platform said on their blog:
Evolving our own container service to the meet the demands of high-powered workloads is another step in advancing the opportunity of containers, as well as designing our cloud for data. Bringing Kubernetes to bare metal will help organizations extract more value out of their data, apps and workloads, and better fuel innovation in areas such as machine learning, which will define the next generation of technology.
Public cloud providers Amazon and Microsoft each offer Kubernetes as a managed service -- AWS Elastic Container service for Kubernetes (EKS) and Azure container service for Kubernetes (AKS), respectively -- but both are available only as a public preview. It is also worth mentioning that neither as yet offers the ability to run a cluster directly on bare metal, which arguably puts IBM with their latest offering one step ahead in a very competitive cloud computing industry.