Torbz - Fotolia
3 Kubernetes platforms for your application needs
Admins must consider their application needs when choosing a container strategy and decide between a full PaaS offering built on containers or a fully compliant Kubernetes service.
Containers have always been a part of platforms such as Pivotal Cloud Foundry and OpenShift. But the sudden rise of Kubernetes has made building applications using containers directly the popular option, and there's a distinct gap between using a full PaaS offering such as Pivotal Cloud Foundry or OpenShift and using a fully compliant Kubernetes platform such as VMware Pivotal Container Service.
Containers have gained a reputation as the building blocks for modern-day microservices application development. As a result, VMware announced its intent to acquire Pivotal in August 2019. But VMware and companies such as Red Hat see Kubernetes very differently.
Red Hat's OpenShift focuses on using Kubernetes as a basis for providing a PaaS offering with open source projects. In comparison, VMware doesn't seem interested in helping customers build PaaS environments outside of its partnership with Pivotal.
As IT administrators look at their container strategies, they must determine what they want to get out of a platform. Pivotal Cloud Foundry (PCF) as a project is moving to Kubernetes and provides containers as a service. OpenShift, which is in beta, also closes the gap between PaaS and container orchestration.
Understanding the different benefits of PCF, VMware's Pivotal Container Service (PKS) and Red Hat's OpenShift helps admins better understand which offering best suits their application needs. Admins must ask themselves whether they require the building blocks of PKS or whether they want a complete PaaS built on containers.
The appeal of the PCF platform
When AWS introduced Lambda, application developers could use it to think through the business logic of a challenge, write code and deploy it to a platform without any consideration of server infrastructure. PaaS offerings such as PCF provide similar capabilities.
However, Lambda focuses on functions as a service (FaaS) versus the entire range of services offered in a complete PaaS offering. For example, Lambda relies on external storage services, while PCF has volume services built in as part of the platform. PCF offers admins an infrastructure that provides all the components -- such as databases, storage, messaging and access control -- required to build most applications and deliver business outcomes.
VMware PKS offers building blocks
VMware's PKS comes in two different offerings: Essential PKS, a quick-deploy model where admins can deliver a simple Kubernetes platform infrastructure; and Enterprise PKS, a heavier deployment for admins further along in their container journeys. VMware designed Enterprise PKS to deploy with the help from professional services.
With PKS, seasoned admins can expose the containers for developers to create the applications they require without the biased services that PCF provides, such as databases and access control. An application team can pull down Istio or NSX Service Mesh for service discovery in PKS, as opposed to Service Registry in PCF.
Admins must select, deploy and manage each portion of a development platform and provide it to a developer community. However, without the structure of a PCF, this process can be overwhelming, especially for admins new to container technology. PKS, like any Kubernetes deployment, functions best as a building block.
OpenShift offers essential PaaS services
OpenShift 4.0 came from IBM's acquisition of Red Hat. In a podcast interview, Brian Gracely, OpenShift's senior director of product strategy, explained that Red Hat built OpenShift on upstream Kubernetes with Red Hat extensions.
OpenShift provides an environment where admins can deploy Kubernetes and create a platform for developers. Red Hat offers a variety of services that make up what's essentially a PaaS. These services include Knative as a FaaS, Istio as a service mesh project, Prometheus as a monitoring system and time series database, and MongoDB as a NoSQL database.