The Technical Oversight Committee (TOC) for the Cloud Native Computing Foundation (CNCF) has elevated the open source Crossplane platform for managing both Kubernetes clusters and legacy IT platforms to become an incubation project.
Incubation is the level CNCF assigns projects before they officially graduate. Crossplane was originally accepted as a sandbox level project to foster development of a universal control plane that enables IT teams to standardize on a single management plane for hybrid cloud computing environments. It was originally developed by Upbound in 2018 and has now been downloaded more than 18 million times, with a 12X increase occurring in the last year.
At its core, Crossplane is an extension of the Kubernetes control plane that makes use of opinionated Crossplane composite resource definitions (XRDs) and Kubernetes custom resource definitions (CRDs) to integrate with legacy platforms in a way that allows Kubernetes operators to be used to manage them.
Upbound CEO Bassam Tabbara says Crossplane makes it possible for operators to compose their own opinionated application programming interfaces (APIs) without having to write any code. IT teams can then present application development teams with a declarative API through which they provision infrastructure. That approach also simplifies operations in a way that eliminates configuration drift, adds Tabbara.
The community is also committing to continue investing in code generation pipelines to expand the surface area of resources supported by Crossplane and its providers. In addition, there will be custom compositions that allow users to express custom logic to generate their platform abstractions using tools of their choice.
Accenture, BBD, CloudCheckr, Deutsche Bahn, Mothership, VSHN, Zego and others are already using Crossplane in production environments. The maintainers of the project are also with all the major cloud providers as well as with other maintainers of CNCF projects. There are now more than 184 Crossplane contributors from 105 companies.
Enterprise IT environments today are challenging to manage because there are so many platforms running a mix of legacy monolithic applications and microservices-based applications that are typically deployed on Kubernetes clusters. IT organizations are now coming under pressure to reduce the total cost of IT by centralizing the management of the extended enterprise.
Organizations that embrace Kubernetes gain a control plane that is embedded within the platform, unlike legacy control planes that might one day need to be extended to support Kubernetes. The path of least resistance for many site reliability engineers (SREs) is to simply extend that framework to other platforms, notes Tabbara. As more Kubernetes clusters are deployed in enterprise IT environments the number of SREs familiar with that construct for managing IT environments continues to increase.
Each organization will naturally need to decide which approach makes the most sense. There is always an upfront cost when implementing any new control plane that needs to be weighed against trying to extend an existing platform or opting to implement multiple control planes that manage platforms in isolation. Regardless of the approach, however, any effort to reduce the total cost of IT needs to start with the control planes employed.