Cloud Native & Kubernetes Trends 2021

Expert Views

Maximilian Hille

Head of Consulting

What do big and powerful trends often bring with them? That’s right, they become more boring when people finally understand them and get on board. Cloud Native has been able to achieve a fabulous and sustained rise in enterprise IT. Application development in and for the cloud, which is what Cloud Native means at its core, has long since escaped the niche. Technologies like Kubernetes, Cloud Native infrastructures or DevOps and business concepts are on the verge of becoming truly mainstream. It is not yet getting boring to watch the rise, deployment and challenges around Cloud Native.

A little over 12 months ago, we laid out the Cloud Native & Kubernetes trends for 2020. Although – or perhaps because – the year turned out to be completely different economically, socially and “pandemically” than most could have predicted, the Cloud Native universe has continued to evolve almost unperturbed, spawning new projects, innovations and opportunities for application development and IT infrastructure. Even in the past, hyperscalers such as Google or AWS have benefited during and through economic crises, showing counter-cyclical and sometimes even erratic growth. Corona is no exception here, as our overall Digital & Technology Trends for 2021 predict.

Before we look at new and upcoming trends, let’s take a quick look at the ones from last year and evaluate the extent to which they raised/lived out to their hype.

You can also skip this chapter and see what the next wave of Cloud Native trends promises!


Trend-Check (Trends 2020 assessed today)

Click here to read the Cloud Native Trends 2020: Link

  • Kubernetes Prod 2.0 – The maturity and capabilities of Kubernetes in production use have increased over the past year. Kubernetes is a driver in automated environments and the applications it runs are becoming increasingly business critical. → on the way
  • Kubernetes-native software boom – More and more solutions are being developed not only Cloud Native, but also directly for use in container environments like Kubernetes – in the enterprise environment and the developer community. → on the way
  • Kubernetes on the Edge – Some projects have been launched in the past year. Edge is a huge trend that somewhat steals the show from Cloud. Kubernetes will play a big role, but needs some more time until reaching maturity. → in the starting blocks for more
  • Kubernetes goes Machine Learning – With KubeFlow & Co. there are projects that are increasingly used to train Machine Learning algorithms. Machine Learning itself is the hype, Kubernetes not a direct enabler, but a companion → No big impact yet, but high expectations.
  • Cloud Native in the data center – Tanzu and the VMware vSphere, Azure Arc, Google Anthos – enough said! Cloud Native comes to the data center and becomes a driver of Hybrid Cloud, even if adoption takes its time. → on the way
  • Kubernetes as a new haven for packaged software – More and more ISVs and software vendors, such as Software AG, are rebuilding their software solutions to become more agile through Kubernetes and containers. → on the way
  • The Secret Winner OpenShift – OpenShift is gaining popularity as Kubernetes deployment and PaaS solutions, especially in hybrid environments (Trend 5). → on the way
  • Hire or Buy or Automate – Enterprises are working meticulously to catch up with the skills gap and get into the driver seat for cloud-native environments themselves, with many initially relying on Kubernetes deployments and managed services. → on the way
  • Moving Beyond Kubernetes – Cloud Native per se means application development and operations in the cloud, for the cloud. Kubernetes and containers have been one of the strongest drivers, but the tools and deployment areas go far beyond the technical requirement of containers. → on the way
  • End Of Docker Supremacy – Docker and Kubernetes have always been independent of each other, just experienced about the same level of adoption. Docker got some takers, was already sold to Mirantis in 2019, and will not be fully supported by Kubernetes going forward. → on the way
  • Microservices with Service Mesh – Service meshes are important to ensure service availability in microservices architectures within containers. Kubernetes relies primarily on Istio for this, but this can be complex to use. Microsoft has promised a new, more open option with the Open Service Mesh, which may soon provide a remedy. → in the starting blocks for more
  • Tracing Boom – Tracing, much like Service Mesh, is a building block necessary for mature container landscapes. It helps ensure that actions in container clusters are traced. More and more projects and services are maturing and being used. → in the starting blocks for more
  • Most Wanted: Standards – There is a consensus that new services and tools are needed to implement Cloud Native in a secure, stable and value-added way for enterprises. The proliferation of these tools makes it difficult to keep track. Standards are strongly desired and are being driven. → on the way
  • Proprietary services vs. open source stacks – The question is far from settled. Standard proprietary services from hyperscalers and cloud-native vendors have their merits when users have chosen a technology stack. Only “real” open source services, which also do not generate any overhead in terms of exit strategy, have the absolute flexibility. → presumably the question will never be answered definitively
  • Kubernetes Provider – ready to rule them all – Kubernetes Service Providers who are knowledgeable in the container technology environment and can implement requirements for enterprise in “brown fields” are the guarantee for success. → on the way


Cloud Native & Kubernetes Trends 2021

Leaving the past behind, let’s take a look at what Cloud Native promises for the future. The following forecasts are to be understood as a concretization of the general digital & technology trends of Cloudflight. regarding the importance of Kubernetes, data in the software stack or NN, some announcements are made that see Cloud Native on the rise again.

Cloud Native and Kubernetes are increasingly becoming an integral part of enterprise IT strategies and are thus also enablers of digital business models – even if only indirectly. However, this will shape development far more than the community would like to admit. For developers, The convenience in handling the technologies is already high. The robustness, integration depth and manageability of the solutions in a complex and heterogeneous stack will therefore be at the top of the agenda for IT decision-makers.

The trends in detail:

  • Consolidation – the M&A wave is rolling – The number of projects, ideas and innovations in the cloud-native universe is shooting through the roof. More and more projects are reaching maturity and are thus ready to be used in enterprise solutions. Numerous other projects currently in incubation are just waiting to be promoted. The call for more clarity and standardization was already loud last year. For this reason the successful projects will separate themselves from the flash in the pan. This means that many large tech companies are putting out feelers for technology and service expertise in the cloud-native environment. They are retooling their product roadmaps and are digging deep just to get a head start. They should be weary about breaking the open source culture as acquisitions might flop because of it. Other projects that are unwanted on the ‘marriage market’, but also not big enough to survive on their own, slowly become less significant.
  • For enterprise use cases, Kubernetes is the “new normal” – Cloud Native architectures are particularly popular when the associated applications are predominantly newly developed. Regardless of the infrastructure basis (private, public, hybrid, multi-cloud, edge or on-premise) as well as the deployment scenarios (web applications, IoT, machine learning), the cloud-native stack seems to have no alternative in terms of flexibility, manageability and developer experience. Thus, cloud-native technologies are giving rise to more and more opportunities to optimize existing applications and decompose them into microservices, as well as to develop new applications natively in Kubernetes. Especially in the enterprise environment, the use of these tools is so widely embraced that Cloud Native has become the new norm. In combination with the PaaS and platform services of the hyperscalers, a powerful new technology ecosystem is emerging that should be cleanly integrated. The warning not to containerize and modify all enterprise workloads indiscriminately in search of perceived performance and flexibility benefits should be taken seriously by decision makers and enterprise developers.
  • Cloud Native also becomes Non-Developer Topic – Cloud Native is polyglot in its definition and application. From a pure technological phenomenon to an all-encompassing design criterion for digitization strategies, everyone can form their own truth. Admittedly, this doesn’t always make it easy to grasp the myth, or to separate hype from reality. Nevertheless, there is no question that Cloud Native is not just a pure phenomenon, but also influences, at least indirectly, the culture and organization of digital value creation in the company. Cloud Native thus attracts far more roles and groups of people than just developers and cloud architects. Even in the former developer hotspots such as GitHub, more and more non-developers are becoming active and driving development around Cloud Native by sharing their requirements, ideas and use cases or even discussing culture and business opportunities.
  • Project overload in the CNCF and the companies – In contradiction to the above-mentioned trends towards standardization in the market and focusing on the company side, it is also becoming apparent that the projects around Kubernetes are becoming more extensive. Due to the increasingly important ecosystem around Kubernetes, which itself comprises only one layer of the architecture (such as logging, security, CI/CD, etc.), new requirements and opportunities are constantly being born. Thus, companies are partly searching for their ideal stack and are still at the mercy of new options. In some cases, developers also build tools and small services themselves, which shortly thereafter become the CNCF’s project. Others deliberately rely on a service provider that offers its architecture as a managed service. Despite lower chances of survival and already great difficulties in structuring, new projects are thus also carried into the companies.
  • Hybrid PaaS thanks to Kubernetes – True multi- and hybrid cloud architectures for application operation have long been the exception. Many companies already have multiple and partially integrated architectures consisting of on-premise, private cloud and multiple public cloud architectures. Through Kubernetes and dedicated services such as Azure Arc, Google Anthos, MongoDB Atlas or Terraform, the clouds are increasingly growing together and are ultimately an abstracted operating environment that specifically accesses microservices and platform services to operate their respective workloads. With increasingly unified technologies that can be deployed regardless of infrastructure environment, enterprises get a gradual risk of a vendor-lock-in at the management level. In theory, thanks to Kubernetes and open source, everything is reversible, but in practice, the new, dynamic application clusters still hang on to the de facto standards of the cloud-native world. In this way, cloud providers subtly succeed in making themselves the No. 1 infrastructure in the company and at the same time play the openness card.
  • Stateful applications discover Kubernetes – The advantages of developing and running applications in container clusters are primarily
    • Resilience – Nodes can be restarted elsewhere for error recovery
    • Immutability – Optimized patching, updates, and rollbacks to reduce errors
    • Portability – Applications can run in other environments independent of the operating system and infrastructure layers
    • Automation – scaling and self-healing

These benefits traditionally fall on so-called stateless (stateless) apps, as these were supported by Kubernetes for various reasons to get started. These stateless apps have no data persistence themselves and always perform every action as if it were the first time. Stateful (conditional) apps, on the other hand, contain data and such as state and context that is critical to their behavior. These include Big Data applications, machine learning or other database types. With new tools and services and a persistent storage layer that can be accessed via API, the usually larger and more enterprise-critical workloads are also supported. Thus, stateful apps in particular will account for a larger share of application growth in cloud-native environments this year, despite somewhat higher complexity.

  • DevSecOps thinks the trend through to the end – It is well known that DevOps and Cloud Native belong together like cinema and popcorn. However, the need to also integrate the topic of IT security directly into the development and quality assurance process is becoming ever stronger. In the agile, flexible microservices environments that can be continuously adapted through CI/CD, there is no longer any room for a separate security cycle that blocks application development in terms of time and composition. Thus, DevSecOps has found a way to make existing and future security challenges an integral part of the development process and toolchain. Technologies such as Runtime Application Self-Protection (RASP) or Software Composition Analysis (SCA) help integrate these methods and also encompass companies’ existing security tools. Through DevSecOps, which in practice is much more than switching a few tools and making some adjustments to the development process, enterprises are becoming sustainably successful with their mission-critical cloud-native workloads. This year, some will venture into DevSecOps, in part due to the push from technology vendors, but true professionalization will likely not be achieved until next year.
  • Serverless is back – on the way to NoOps – For some time now, the topic of serverless has been hotly debated in the environment of cloud architectures. Even if the decisive advantages with the independence of the underlying infrastructure and the exclusively event-based computing are only designed for certain use cases with very selective load scenarios, the trend is spreading more and more. Serverless is also viewed critically because it puts companies in a lock-in at a very high level and makes them less flexible in terms of programming. However, one driver of Serverless or NoOps is also container technology. Due to the property of portability, the underlying infrastructure only plays a subordinate role there. Container orchestration with Kubernetes ensures that the necessary computing resources are always provided. Thanks to automation and independence, applications are basically not influenced by the operating environment. AWS in particular, which has long been on the serverless bandwagon with Lambda, announced some products and services at its re:invent in November/December that highlight the intersection of containers and serverless. This means that NoOps application landscapes will play a role at many companies this year.
  • AIOps becomes reality – By using Kubernetes in application operations, companies have already been able to significantly gain speed, flexibility and innovation for digital business. In the right setting, Kubernetes clusters can be largely automated and “self-healing”. However, in terms of cost and stability, enterprises and developers are still ultimately on their own. The right sizing of the architecture, the appropriate use of tools, transparency about incidents and much more is the responsibility of the users or their service providers. With services based on Kubernetes, which additionally evaluate all existing data in operation (such as performance, failures or anomalies) and transfer them to the control of the container clusters, these worries are also relieved. Proper training of the algorithms for the appropriate application operation is critical. After all, if the clusters’ automation commands are stored incorrectly, major damage can quickly result. Distrust of the technologies, lack of experience and criticality in terms of container volume have tended to make AIOps the north star of application operations. However, they are already more common in modern and elaborate setups today. Therefore, AIOps will become more of a standard this year and, with the help of appropriate tools or blueprint stacks, an elementary part of the cloud architecture.
  • CRI-O becomes the new Docker – Already last year we predicted the end of Docker supremacy. The myth that Docker and Kubernetes are a covenant for life and inseparable was “busted”. The limitations that containers, as a monolith among lightweight runtimes, brought were overcome by other concepts. The CRI-O project, initiated by Red Hat, is one of several alternatives for Docker that works as a container runtime. While the architectures are variable and open, CRI-O seems to have secured momentum at the moment, which may be due in no small part to its proximity to the OpenShift platform, which is also riding high alongside Plain Kubernetes and hyperscaler distributions. There will be no monopoly on container runtimes in the future. Nevertheless, CRI-O will clearly gain momentum this year.

The growth in popularity and adoption of Cloud Native and especially Kubernetes-based architectures seems to continue unabated in 2021. The deployment scenarios and possibilities of cloud-native technologies are increasingly convincing to corporate decision-makers. With Cloud Native and Kubernetes, the digital strategy and modernization of the IT landscape will still not be a no-brainer, but it seems that it will no longer work without them either.

Companies will only be successful if they define their very own and foresighted transition and future strategy. It is not always right to indiscriminately modernize all applications, migrate them to the cloud, break them down to microservices and package them in containers. They should rather identify the right and decisive use cases in a targeted manner. A service and fulfillment partner can help with this.

The trends described above, as well as the roadmap of the CNCF and its partners with the numerous ongoing projects, should help to further increase stability, relevance and thus business success on the cloud. It is to be hoped that users will not lose their heads this year and will keep track of things.


Study on the topic:

In addition to our Cloud Native Trends, Cloudflight has also empirically investigated the current development status and opportunities of Cloud Native in enterprise use. Our latest study “The Rise of Cloud Native” is available for free download at this link.