This is the third post in Seven Major Cloud Native Trends for 2020 Series
In 2019, with the continuous progress of the community’s top participants, “scale” and “performance” finally became important buzzwords in the Kubernetes project. This not only truly went the last mile in achieving Kubernetes’ large-scale implementation in enterprise production environments, but also made Kubernetes a real technological leader in high-level and large-scale Internet scenarios, such as “Double 11”, for the first time.
As we enter the new decade, we think the Kubernetes field will undergo the following changes in 2020:
Kubernetes Will Become a New Interactive Interface between Users and Cloud Computing
With the popularization of cloud native computing, more and more application loads will be deployed on Kubernetes, including databases, big data, AI, and innovative applications. Kubernetes has become the cornerstone of cloud native computing. Thanks to the large-scale application management capabilities of Kubernetes and its ability to support hybrid cloud scenarios, Kubernetes will become a new interactive interface between users and cloud computing in 2020. From the perspective of architecture, Kubernetes has become the control plane of the IaaS layer and promoted capability optimization for the underlying IaaS (computing, storage, and network) to meet containers’ needs for high density and high dynamism.
Controlling Kubernetes has Become a Core Skill of Enterprise O&M Teams, and This Promotes Mutual Development with AIOps
On the other hand, will the extensive application of Kubernetes lead to mass layoffs of enterprise O&M personnel? In fact, the proliferation of enterprise IT architectures, the transition from on-Kubernetes to in-Kubernetes, and the introduction of a large number of CRDs, custom controllers, and service meshes pose major challenges to Kubernetes stability and performance optimization. Kubernetes mastery has gradually become an important assessment criterion for the technical capabilities of enterprise O&M teams. The skills of enterprise O&M personnel are evolving from automation to digitization and intelligence.
In 2020, we expect that, in the Kubernetes field, AIOps will gradually emerge to further improve Kubernetes cost optimization, fault detection, and cluster optimization.
At the same time, cloud native technologies such as Kubernetes, will give AIOps a clear path forward:
1) Due to the sound design of Kubernetes, including declarative APIs, immutable architecture, and elegant scaling mechanisms, it can promote the normalization of application release and O&M operations.
2) By implementing automated processes, such as GItOps, Tekton, and SecOps, the application lifecycle will be further standardized.
3) With the development of projects such as OpenTelemetry and CloudEvents, the application observability field will be further standardized and integrated in terms of logs, monitoring, tracing, and events. This will enrich the datasets for multiple metrics and root cause analysis, improving the AI accuracy and coverage of AIOps.
New Kernels and Hardware Will Boost the Evolution of Container-Optimized Operating Systems
After years of development, from the early Docker, rkt, and CRI-O to containerd, Kata Container, and gVisor, container technology has been a critical foundation for empowering Kubernetes. However, continuous polishing and enhancement are required for the further isolation of runc scenarios and for better performance in security container scenarios.
As new kernel technologies, such as CGroup V2, namespace, and virtiofs, become more mature, container runtime capabilities can be further enhanced. On the other hand, the introduction of new hardware, such as NPU, MoC, and NUMA, also allows for more optimizations in container and Kubernetes scheduling. By incorporating these capabilities, container-optimized operating systems tailored to container scenarios became a possibility and then developed rapidly.
The Container Network and the Mesh Network Will Be further Integrated
After years of market cultivation, 2020 will be the year service mesh achieves widespread popularity. In the meantime, service mesh performance optimization will become the main challenge. Some sinking solutions are further optimizing network forwarding performance based on container network interface (CNI) and kernel technology.
The container network itself is also evolving, from IP-oriented to identity-oriented and from a single container network plane to a multi-network plane. This further optimizes network forwarding performance and zero-trust security. In 2020, we believe that the container network and the mesh network will further fuse and be integrated in Network ServiceMesh, NFV, and other scenarios.
You May Also Like
In this blog, an Alibaba engineer discusses his thoughts on DevOps, cloud, and Kubernetes, and addresses whether Kubernetes will kill off DevOps or not.
This article explores the reasons why 2019 stood out as the critical year of the cloud-native era and gives a brief account of the history of cloud-native technologies.