Alibaba Cloud's ACK One multi-cluster management has been fully upgraded: multi-cluster services, multi-cluster monitoring, application disaster recovery in two places and three centers

Author: Yuhui

ACK One overview

ACK One is a distributed cloud container platform launched by Alibaba Cloud for scenarios such as hybrid cloud, multi-cluster, and distributed computing. It can manage Kubernetes clusters on Alibaba Cloud, at the edge, and deployed in customer data centers and other clouds in a unified manner, and simplify clusters. The management interface allows enterprises to easily deal with multi-cluster forms, so as to flexibly meet their own business and data management and control needs, and at the same time enable clusters deployed in customer data centers to obtain elastic computing power on the cloud, realizing "computing without boundaries".

Three major features, covering three scenarios, leading to a new height of multi-cluster management

Multi-cluster service--cross-cluster service discovery and access

• Overview

Using Kubernetes' native Service, you can easily discover and access services within the cluster, but you cannot realize cross-cluster service discovery and access. With the development of business, applications often need multi-cluster deployment to achieve isolation and high availability, such as: multiple development and production clusters, different clusters for different businesses, clusters maintained by different organizations, to solve availability and data ownership requirements, flexible account sharing, etc. need. At the same time, different business applications need to access each other, so a cross-cluster service solution is needed to deal with the challenges brought by multiple clusters.

ACK One multi-cluster management releases a new feature multi-cluster service, and realizes cross-cluster access to Kubernetes native services through multi-cluster services, breaking the boundaries of multi-cluster service access.

• Scenario 1: Cross-cluster service discovery and access

ACK One multi-cluster service supports the Kubernetes community multi-cluster service API standard, and realizes cross-cluster service discovery and access through ServiceExport and ServiceImport objects.

1. As shown in link 0 in the figure, the administrator delivers resources such as applications, multi-cluster services ServiceExport, ServiceImport, and distribution rules through the master instance.
2. As shown in link 1 in the figure, the master instance delivers application and multi-cluster service resources to the associated clusters according to the distribution rules.

a. By issuing ServiceExport resources to ACK Cluster 1, define ACK Cluster 1 as a service provider.

b. By delivering the ServiceImport resource to ACK Cluster 2, define ACK Cluster 2 as a service consumer.

3. As shown in link 2 in the figure: Client Pod on ACK Cluster 2 can access Service 1 on ACK Cluster 1.

• Scenario 2: High availability of multi-cluster services

1. As shown in link 0 in the figure, the administrator delivers resources such as applications, multi-cluster services ServiceExport, ServiceImport, and distribution rules through the master instance.
2. As shown in link 1 in the figure:

a. Through the ACK One master instance, deploy the service Service1 in ACK Cluster 1 and ACK Cluster 3 at the same time to achieve high availability deployment.

b. By issuing ServiceExport resources to ACK Cluster 1 and ACK Cluster 3, define ACK Cluster 1 and ACK Cluster 3 as service providers.

c. By delivering the ServiceImport resource to ACK Cluster 2, define ACK Cluster 2 as a service consumer.

3. As shown in link 2 in the figure: Client Pod on ACK Cluster 2 can access Service 1 on ACK Cluster 1 and ACK Cluster 3 in a load-balanced manner.

• Advantages of multi-cluster service solutions

Compared with the traditional cross-cluster exposure schemes of Kubernetes services such as LoadBalance/Ingress, ACK One multi-cluster service does not need to introduce additional load balancing resources. It can realize cross-cluster service discovery and access only by defining service providers and consumers.

Compared with the multi-cluster service solution of the service mesh, the implementation of the ACK One multi-cluster service is very lightweight, does not require the introduction of sidecar containers, and has no additional network delay. At the same time, the ACK One multi-cluster service is hosted on the ACK One control plane, saving O&M costs.

For details, please refer to the end of the article: Multi-cluster service.

Multi-cluster global monitoring--a tool for operation and maintenance, global monitoring of the market

• Overview

Prometheus monitoring is the de facto standard for Kubernetes cluster monitoring. Prometheus monitoring components can be installed in the ACK container service to quickly implement ACK cluster monitoring. The ACK container service also provides a large number of monitoring panels for use. But in a multi-cluster scenario, users have to maintain the Prometheus monitoring component of each cluster separately. In daily operation and maintenance, it is necessary to frequently switch the monitoring panels of different clusters. In problem diagnosis, it is necessary to manually compare the differences in the monitoring indicators of each cluster.

ACK One multi-cluster global monitoring is based on the monitoring indicators of single-cluster Prometheus, aggregates monitoring indicators of multiple clusters through multi-cluster monitoring aggregation instances, and provides a global monitoring dashboard from a multi-cluster perspective, allowing you to obtain multiple monitoring dashboards at the same time. The monitoring indicators of the cluster are convenient for daily operation and maintenance work.

• Global monitoring of the market

Currently supported large disks include: global cluster stateless application monitoring, TopN global cluster node monitoring, and TopN global cluster Pod monitoring. At the same time, ACK One global monitoring supports custom monitoring dashboards, and you can build your own multi-cluster monitoring dashboards through aggregated instances.

For details, see the end of this article: Multi-cluster global monitoring.

Multi-cluster, two-site, three-center application deployment -- a new choice for application disaster recovery

• Overview

Three centers in two places refers to the deployment of three business processing centers in two cities, namely: a production center, a disaster recovery center in the same city, and a disaster recovery center in a different place. Deploy 2 sets of environments in a city to form dual centers in the same city, process services at the same time and achieve data synchronization through high-speed links, which can be switched to operate. Deploy a set of environments in another city as an off-site disaster recovery center for data backup. When both centers fail at the same time, the off-site disaster recovery center can switch to process services. The disaster recovery solution of three centers in two places can guarantee the continuous operation of business to a great extent.

• Scheme Architecture

Using ACK One's multi-cluster management application distribution function can help enterprises manage 3 K8s clusters in a unified manner, realize rapid deployment and upgrade of applications in 3 K8s clusters, and achieve differentiated configuration of applications in 3 K8s clusters. Combined with GTM (Global Traffic Management), service traffic can be automatically switched between three K8s clusters when a fault occurs. For data replication at the RDS data plane, refer to DTS data transmission service.

For details, please refer to the end of the article: Using ACK One to build a disaster recovery solution for two locations and three centers for application systems.

Summarize

This article introduces the three main features recently released by ACK One, covering three main scenarios of multi-cluster management, cross-cluster service discovery and access, multi-cluster global monitoring, and application disaster recovery. In addition to multi-cluster management, ACK One supports connecting and managing Kubernetes clusters in any region and on any infrastructure, provides consistent management and community-compatible APIs, and supports computing, networking, storage, security, monitoring, logging, and jobs. , applications, traffic, etc. for unified operation and maintenance management and control. Alibaba Cloud Distributed Cloud Container Platform (ACK One for short) is an enterprise-level cloud-native platform for scenarios such as hybrid cloud, multi-cluster, distributed computing, and disaster recovery.

The distributed cloud container platform ACK One product is in the free public beta, please click here to go to the ACK One product details page to open the experience.

Reference link:

[1] ACK One, a distributed cloud container platform:

https://www.aliyun.com/product/aliware/adcp

*[2] ACK One Product Console:*

https://cs.console.aliyun.com/one

*[3] Multi-cluster service:*

https://help.aliyun.com/document_detail/440680.html

*[4] Multi-cluster global monitoring:*

https://help.aliyun.com/document_detail/434106.html

*[ 5] Multi-cluster application delivery:*

https://help.aliyun.com/document_detail/419336.html

*[6] Using ACK One to build a disaster recovery solution for two locations and three centers for application systems:*

https://help.aliyun.com/document_detail/427364.html

If you want to know more about ACK One, welcome to Dingding search group number: 35688562 to join the group communication.