As we look ahead to 2019, we take a view of the industry of Kubernetes, containers, and cloud platforms. Here are some predictions on what we see ahead.
1. Kubernetes and containers are widely understood. Now it’s time for automated operations, agnostic of the underlying cloud provider.
Having been open sourced in 2014 and used in production by enterprise teams today, Kubernetes is now better understood. We are entering the third era of Kubernetes, meaning users will be looking at ways to bring in more automation around operations. Operation teams will be looking for the qualities typically associated with hosted services, such as automated updates, automated back-ups, auto-scaling and self-tuning, to be available on any environment, whether on a cloud provider’s infrastructure, or on their own premises.
In 2019, more and more, the automation of these operations will manifest themselves as Operators: Operators take human operational knowledge of a given application or service and encode it into software. They help to codify operational processes into Kubernetes-native infrastructure and services running on top of it, providing a more efficient way of managing Kubernetes-native applications at scale. What’s more, this codification will be implemented by subject matter experts with deep hands-on experience operating these infrastructures and services themselves.
We have already started seeing this trend with OpenShift 3.11, where we introduced a first round of Operators for MongoDB, Redis, Crunchy Data for PostgreSQL, Couchbase, etcd, and Prometheus. These operators encapsulate years of hard earned operational experience from the vendors and experts behind each. In 2019, we are going to see this paradigm expand across the Kubernetes landscape, with Red Hat OpenShift 4.0 leading the way, and across hundreds of diverse services which, thanks to their Operators, can provide cloud like automated operations across infrastructures.
The increasing shift towards automated operations represents a third wave in container native infrastructure services, allowing teams to begin thinking in terms of deploying third party tools across their clusters in one move, rather than slowly rolling out services in steps as they are scaled through configuration and systems customization over time.
2. Federation V2 will make multi-cloud easier.
The latest work on Federation inside the Cloud Native Computing Foundation (CNCF) has yielded Federation V2, which addresses many of the use case challenges that have arisen as the first few waves of cluster orchestration problems have given way to the next layer of challenges. One of those challenges was in federating identities and workload across clusters, regardless of geographic location or underlying infrastructure. As Federation V2 matures and is adopted, the ability to operate multiple clusters across differing infrastructures and cloud providers should be made easier.
3. Serverless comes to Kubernetes, and 2019 will be the year of Hybrid Serverless.
This is the year where serverless, specifically Functions-as-a-Service (FaaS), breaks free from the tentacles of a single cloud provider and goes deep and wide. It goes wide since developers will be able to not only use their FaaS of choice on their cloud provider of choice, but will also be able to expand the serverless paradigm beyond functions and to their favorite stack: e.g. a spring boot API implementation can be serverless-enabled so that it executes only on its API calls and scales to zero otherwise. Going deeper, the events source behind serverless functions and applications will expand to a heterogenous ecosystem spawning the diverse range of services out there, as opposed to the limited set of proprietary cloud provider services that are available for serverless applications today.
While FaaS has become a more standard tool in the developer toolbox, it’s still currently the case that each serverless environment is essentially an island unto itself. That wild divergence between platforms should begin to abate, thanks to the introduction of Knative, the project which allows for serverless computing to be executed within Kubernetes.
4. Kubernetes will enable hybrid operations for containers AND virtual machines...and it’ll be on bare metal.
We’ve previously thought of virtual machines as the “old world” and containers and Kubernetes-native applications as the “new.” 2019 will see that mindset change, as the advent of projects like Container-native Virtualization (enabled by KubeVirt) make the choice between VM-centric and Kubernetes-centric infrastructure moot.
As Kubernetes takes hold in enterprises, it provides a more flexible, scalable model for production workloads, but those running in Linux containers. In 2015-2018, this meant that Kubernetes was primarily for greenfield or re-architected applications. That is no longer the case with Container-native Virtualization, which enables virtual machines to follow the same workflow as Kubernetes-native applications. By breaking down the wall between old and new operations, enterprises will be able to more effectively integrate operations and retain existing IT skills while still embracing modern infrastructure built on Kubernetes.
Additionally, this parity of innovation between virtual machines and Linux containers will be ready for bare-metal servers. While complicated virtualization stacks had been necessary for modern infrastructure in the past, advances in how Kubernetes runs on bare metal will enable organizations to take full advantage of the greater speed and efficiencies posed by deploying Kubernetes-native infrastructure directly on servers.
5. Open source developer communities target Kubernetes by default.
As Kubernetes has become readily available on every major cloud platform we are just beginning to see the emergence of open source developers targeting their projects to Kubernetes first. While most of 2018 saw Kubernetes growing fast and seeing numerous new contributors and projects, 2019 will likely see the results of that consolidation in the form of a wildly expanded ecosystem around Kubernetes. Examples of developer communities targeting Kubernetes already include Trillian, Source Graph and GraphQL (Apollo and Hasura).
6. We will see some workloads start moving back to the datacenter.
Developers will be able to leverage a more and more diverse set of services, from a wide ecosystem of communities and vendors, backed by automated operations that make them as accessible and easy to operate as any cloud service. This trend, coupled with increasing occurrences of sticker shock from cloud provider infrastructure bills, will make CIOs think more carefully about their cloud workloads, and even in some cases, take select workloads that are running on cloud providers infrastructure back to on-premises to decrease costs while being able to maintain the automation benefits. Thankfully, Kubernetes-native infrastructure enables CIOs to optimize their use of cloud services and resources and still achieve service availability/resilience and security objectives.
In summary, looking at what is to come around Kubernetes and the hybrid cloud ecosystem, we are excited for what’s ahead. As the developers of Red Hat OpenShift, the Kubernetes platform for big ideas, we’ve worked with the early adopters that have already begun delivering their next generation applications. Now, it’s time for many more to cross the chasm to use Kubernetes, Operators, and beyond to begin reaching new levels of innovation for 2019.