The evolution of Kubernetes clusters in multi-cloud and hybrid cloud

As a core technology, Kubernetes has become the foundation of modern application architecture, and more and more enterprises use Kubernetes as the container orchestration system.

The following data comes from the raw data of the 2022 CNCF Survey. It can be seen that the proportion of enterprises using Kubernetes has reached 80%.

The popularity of Kubernetes is mainly due to the following reasons:

• Automation: Kubernetes automates operations such as container deployment, expansion, load balancing, failure recovery, and rolling updates, greatly simplifying application management and maintenance. This automation also improves application resiliency and availability.
• Portability: Kubernetes’ container-based architectural model allows applications to run on any cloud platform, physical machine, or virtual machine without recoding or updating configurations.
• Ecosystem: As a successful open source project, Kubernetes has strong community support and ecosystem, enabling it to obtain better innovation, optimization and security guarantees. From the community, we can find various plugins and tools, providing developers with rich choices and extensibility.

From a single Kubernetes cluster to multiple Kubernetes clusters

Introduced

Applications in enterprises are often complex and require different environments for development, testing, and production deployment. To avoid interference and crossover between applications, it is often necessary to deploy and manage applications separately in different Kubernetes clusters.

After deploying independent Kubernetes clusters in different environments in the same data center, the cluster scale, management methods, reliability, and security in different environments are different. From development, testing to production, the cost investment is also gradually increased to ensure Better performance, increased reliability and security.

This is also a form of multi-Kubernetes clusters (note that multi-Kubernetes clusters are mentioned here).

Rapid development

With the increasing acceptance of cloud computing and the continuous growth of enterprise scale, more and more enterprises have begun to consider adopting or have adopted multi-cloud and hybrid cloud architectures. There are many driving factors for multi-cloud and hybrid cloud, which can be divided into two categories after summarization: active factors and passive factors.

Active Factors

• Avoid vendor lock-in: Avoid vendor lock-in: By adopting a multi-cloud hybrid cloud strategy, enterprises can avoid excessive reliance on a single cloud service provider, thereby mitigating the risks brought about by vendor lock-in.
• Improve performance and reduce latency: According to business needs, enterprises deploy services in data centers that are geographically closer to users.
• Wider range of scalability: Realize elastic scaling of resources between different cloud service providers, thereby providing a wider range of scalability to meet business needs.
• Improve reliability: After application deployment between different cloud service providers and private data centers, even if a cloud service provider fails, enterprise applications can continue to run.
• Cost factor: The services provided by various cloud service providers can be selectively used in order to find the best balance between cost and performance.

Passive Factors

• Data center capacity limitation: The scale and performance of the data center infrastructure cannot meet the demand.
• The transition period between cloud migration and cloud migration: During the process of cloud migration or migration, there is a phenomenon of coexistence of public cloud and private cloud, and the transition period may be prolonged.
• Limited single cloud coverage area: Although cloud service providers continue to expand their global data center coverage, there may still be situations where a single cloud service provider cannot meet the deployment needs of enterprises in specific regions.
• Business isolation: In some cases, enterprises need to isolate different businesses to reduce risks and improve security. Deploy applications to different data centers to ensure the security of sensitive data and critical applications.
• Compliance and data segregation: Different countries and regions may have different regulatory requirements, which may restrict enterprises from storing and processing data on specific cloud service providers.
• With the introduction of multi-cloud and hybrid cloud strategies, the number of Kubernetes clusters is also increasing accordingly. Although the number of clusters is increasing, the essence is the same as a single cluster: each Kubernetes cluster has its own control plane (api-server, controller) and a set of working nodes, which can independently deploy and manage applications (this is Why is it called multiple Kubernetes clusters earlier). In a “multi-cluster” environment, each cluster is independent, and when it needs to be managed, it adopts the “management one by one” method.

A lot of challenges

Multiple Kubernetes clusters can be deployed on different cloud platforms, different data centers, different network environments, and different physical infrastructures to meet different application and business needs. However, with the increase in the number of clusters, there are also many challenges:

• Increased cluster management complexity: Different Kubernetes clusters are deployed on different cloud service provider platforms and have different network architectures and security policies. As the number increases, so does the complexity of cluster management. This includes aspects such as resource allocation, access control, network management, cluster upgrades and maintenance.
• Increased application management costs: Enterprises need to deploy and manage applications on multiple Kubernetes clusters.
• Cross-cluster network and service communication: Kubernetes has the natural network isolation feature, and because it is deployed on cloud platforms with different network architectures, it is necessary to solve the problem of cross-cluster network and service communication. This includes issues such as cross-platform network interconnection, cross-cluster service discovery and routing.

From multiple Kubernetes clusters to Kubernetes multiple clusters

In order to cope with the above challenges, another concept Kubernetes multi-cluster has emerged. Kubernetes multi-clusters and multi-Kubernetes clusters are two different concepts. To facilitate the distinction, we can also call them Kubernetes federations.

Kubernetes federation is to connect multiple Kubernetes clusters together to work together and realize unified management of cross-cluster resources and applications, cross-region and cross-cloud failover, dynamically allocate resources according to actual needs, reduce costs, improve business flexibility and scalability.

Although there is no official solution for how to connect multiple Kubernetes clusters, thanks to the strong community and ecology, there are many open source and commercial solutions.

Cluster Management Complexity

Use a centralized multi-cluster management platform to manage Kubernetes clusters in multi-cloud hybrid clouds in a unified interface, provide cluster creation, configuration, monitoring and troubleshooting functions, make cluster management easier and more efficient, and improve cluster reliability and availability.

In terms of tools, there are KubeSphere, Rancher, and many public cloud solutions such as Google Anthos, Azure Arc, Red Hat Advanced Cluster Management (ACM), AliCloud ACK, etc.

Application Management

Use an automated multi-cluster application orchestration and management platform to solve the problems of rapid deployment, expansion and management of applications in multi-cluster and multi-cloud environments, and ensure the reliability of applications.

In addition to the aforementioned multi-cluster management platform that provides application management capabilities, there are also open source multi-cluster application management platforms such as Kubernetes Federation (the former has been archived in August 2022), and Karmada.

Cross-cluster network and service communication

Reduce network complexity in multi-cluster environments. Use service grid technology to manage service traffic, security and observability across multiple clusters, and realize cross-cluster service discovery and communication; or use cross-cluster network solutions to achieve network interoperability.

Service mesh technologies include Istio, Linkerd, Flomesh, etc., and network solutions include Submariner.

Summarize

At the moment when multi-cloud and hybrid cloud are becoming the trend, Kubernetes multi-cluster is constantly being mentioned and has gradually become an important solution. The discussion of Kubernetes multi-cluster is no longer limited to cluster and application management, but also introduces cross-cluster service communication.

Multi-cluster can provide enterprises with more flexible and reliable application deployment and management capabilities, enabling enterprises to better utilize resources in a hybrid multi-cloud environment, reduce operating costs, and improve application reliability. In addition, it can also help enterprises to perform application migration more quickly.