All Products
Search
Document Center

Container Service for Kubernetes:Historical release notes (2020 and before)

Last Updated:Mar 19, 2025

This topic describes the historical release notes for Container Service for Kubernetes (ACK) features that were released in 2020 and before, and provides links to the relevant references.

Background information

  • The following Kubernetes versions are supported by ACK: 1.24, 1.22, and 1.20.

  • The following operating systems are supported by ACK: CentOS 7.9, Alibaba Cloud Linux 3.2104, Alibaba Cloud Linux 2.1903, Windows Server 2019, and Windows Server Core 1909.

December 2020

Feature

Description

Region

References

New region

ACK is available in the China (Guangzhou) region.

China (Guangzhou)

Quotas and limits

Hot migration from ACK standard clusters to ACK Pro clusters

Hot migration from existing ACK standard clusters to ACK Pro clusters is supported. Your services are not affected during the migration.

ACK Pro clusters are developed based on ACK managed clusters. This type of cluster provides higher reliability and security in large-scale production environments for enterprise users. ACK Pro clusters are also covered by service level agreements (SLAs) that include compensation clauses.

All regions

Hot migration from ACK standard clusters to ACK Pro clusters

SLB specification selectable for the API server

The specification of the SLB instance that is used to access the API server can be selected when you create an ACK cluster. You can select different SLB specifications based on your business requirements. This allows you to handle different traffic loads on the API server of the cluster.

All regions

Create an ACK Pro cluster

Preemptible instances supported by node pools

Preemptible instances are supported when you set the billing method of a node pool. Preemptible instances are cost-effective. You can bid for idle resources of Alibaba Cloud, obtain the resources, and then run containers until the resources are reclaimed due to higher bids from other customers. This reduces the cost of computing resources.

All regions

No impact on workloads

Kubernetes 1.18

Updates from Kubernetes 1.16 to 1.18 are supported.

All regions

Update the Kubernetes version of an ACK cluster

CronHPA

Cron Horizontal Pod Autoscaler (CronHPA) can be enabled in the ACK console for your workloads. You must install ack-kubernetes-cronhpa-controller in the cluster before you enable CronHPA.

All regions

Use CronHPA for scheduled horizontal scaling

CentOS 7.8

CentOS 7.8 can be used as the node OS when you create a cluster or a node pool.

All regions

Create and manage a node pool

Reinforcement based on classified protection

Reinforcement based on classified protection is supported for the cloud-native Alibaba Cloud Linux operating system in compliance with Multi-Level Protection Scheme (MLPS) 2.0 level 3 standards. The following features are provided:

  • Implement identity authentication

  • Access control

  • Security auditing

  • Intrusion prevention

  • Malicious code protection

To enable reinforcement based on classified protection for the node OS when you create a cluster or a node pool, you must select Alibaba Cloud Linux 2.1903 as the node OS and select Reinforcement based on classified protection.

All regions

Volume snapshots supported by CSI

Volume snapshots created from disks are supported by the CSI component of ACK. This allows you to back up and restore workload data.

All regions

Create a snapshot of a disk volume

Cluster upgrade and new components

ACK Serverless clusters can be upgraded. The metrics-server, cronhpa-controller, and alb-ingress-controller components can be installed and managed on the Add-ons page of the ACK console.

All regions

No impact on workloads

November 2020

Feature

Description

Region

References

Managed node pools

Managed node pools that are provided by ACK support auto upgrade and auto repair. This provides centralized, managed, and O&M-free lifecycle management of nodes. You do not need to be concerned about the O&M of nodes, such as component upgrading, OS upgrading, or patching to fix CVE vulnerabilities. ACK automatically fixes node exceptions for the nodes in a managed node pool.

Managed node pools are supported by ACK Pro clusters.

All regions

Managed node pool overview

Kubernetes 1.18 supported by kubernetes-dashboard

Kubernetes 1.18 is supported by the kubernetes-dashboard application provided by App Catalog. This fixes the issue that the pods of Kubernetes 1.18 cannot be accessed by terminals. You can find and install the Helm chart for kubernetes-dashboard from App Catalog.

All regions

App Marketplace

Performance levels of ESSDs configurable

The performance level of an ESSD can be set to PL0 or PL1 when you create a cluster. This allows you to customize the performance level of your cluster.

This feature is supported by ACK Pro clusters, ACK standard clusters, ACK dedicated clusters, and ACK edge clusters.

All regions

FAQ about Elastic Block Storage

CCM

The CCM is updated to V1.9.3.339-g9830b58-aliyun. Hash values are supported in the configurations of LoadBalancer Services. This way, when the CCM is restarted, only the backend vServer groups of the related SLB instances are updated if the Service configuration is not changed. The configurations of the related SLB instances and listeners are not updated.

All regions

Cloud Controller Manager

Disk monitoring supported by CSI

Disk monitoring is supported by the latest version of the CSI component. This feature allows you to monitor the status of persistent volume claims (PVCs) through Prometheus Service when you use disks that are mounted by using the PVCs. You can also configure alerts by setting thresholds for the storage space and input/output operations per second (IOPS) of the disks.

All regions

No impact on workloads

Ingress controller and CoreDNS

Ingress controllers and CoreDNS can be installed when you create an ACK Serverless cluster. You can also install CoreDNS on the Add-ons page of the ACK console after the cluster is created.

All regions

Create an ACK Serverless cluster

Node pools supported by registered clusters

Node pools are supported in registered clusters. You can use a node pool in the ACK console to manage a set of ECS instances for a registered cluster. You can add ECS nodes from a node pool to a self-managed Kubernetes cluster or a Kubernetes cluster that is deployed in the public cloud of a third-party cloud service provider. You can also use node pools to manage the labels and taints of nodes in node pools.

All regions

Create and manage a node pool

October 2020

Feature

Description

Region

References

Time zone

The time zone can be selected when you create a cluster. By default, the time zone of your browser is selected.

This feature is supported by ACK Pro clusters, ACK standard clusters, ACK dedicated clusters, and ACK Serverless clusters.

All regions

Create an ACK managed cluster

Tagging supported for cloud disks, NAS file systems, and Log Service projects

Cloud disks, NAS file systems, and Log Service projects with tags are supported by CSI and Logtail. Cloud disks, NAS file systems, and Log Service projects that are created by ACK for a cluster are added with the cluster ID as tags. This makes it easier to allocate resource fees.

All regions

No impact on workloads

September 2020

Feature

Description

Region

References

New region

ACK is available in the China (Ulanqab) region.

All regions

Overview of ACK Pro clusters

SMB supported by Windows containers

Server Message Block (SMB) file systems can be mounted to a Windows container. In the NAS console, you can create an SMB file system in the VPC where the cluster is deployed. You can also create a mount target for the file system. You must use the FlexVolume plug-in to mount an SMB file system.

All regions

None

Time zone

The time zone can be selected for master nodes and worker nodes when you create an ACK dedicated cluster or an ACK managed cluster.

All regions

No impact on workloads

Kubernetes 1.18

Kubernetes 1.18.8 is supported. You can select this Kubernetes version when you create a cluster.

ACK clusters of Kubernetes 1.18 or later no longer support Kubernetes Dashboard. To use Kubernetes Dashboard, we recommend that you install kubernetes-dashboard on the App Catalog page.

All regions

(Discontinued) Kubernetes 1.18 release notes and [Product Changes] ACK ends support for Kubernetes Dashboard

NetworkPolicy configurable for Terway

The NetworkPolicy feature can be enabled or disabled for Terway when you create a cluster.

All regions

Periodic inspection

Periodic inspection policies can be configured for a cluster on the Inspections page of the ACK console.

All regions

Use the inspection feature to detect security risks in the workloads of an ACK cluster

Cluster auditing

The cluster auditing feature can be enabled or disabled on the Cluster Auditing page of the ACK console.

All regions

Work with cluster auditing

New components

The logtail-ds component is provided to collect container log from registered clusters, including stdout and log files of containers.

The migrate-controller component is provided to migrate applications across Kubernetes clusters. This component is developed based on the open source Velero project.

The ack-virtual-node component is provided to enable auto scaling for registered clusters.

All regions

Sandboxed-Container 2.0

Sandboxed-Container is updated to V2.0. Sandboxed-Container 2.0 has the following benefits:

  • Sandboxed-Container is a container runtime that is developed by Alibaba Cloud based on lightweight virtual machines. Compared with Sandboxed-Container 1.0, Sandboxed-Container 2.0 supports more lightweight and efficient deployment and simplifies the architecture and maintenance of ACK clusters.

  • Sandboxed-Container 2.0 reduces the resource overheads by 90% and improves the startup speed of sandboxed containers by three times.

  • Sandboxed-Container 2.0 increases the deployment density of sandboxed containers on a single node by 10 times.

  • Sandboxed-Container 2.0 supports the virtio-fs file system, which provides higher performance than the 9pfs file system.

All regions

Overview

Knative component supported by ACK Serverless clusters

Knative components are supported in ACK Serverless clusters. Knative is a cloud-native and cross-platform orchestration engine for serverless applications. You can deploy Knative in ACK Serverless clusters. This allows you to use cloud resources by calling the Knative API without the need to pay for the Knative controller.

All regions

Knative

August 2020

Feature

Description

Region

References

Gatekeeper

The gatekeeper component can be installed on the Add-ons page of the ACK console. This component facilitates the management and implementation of policies that are executed by OPA in ACK clusters.

All regions

gatekeeper

Runtime inspection

Runtime inspections can be performed on the Runtime Security page of the ACK console. This feature monitors the container runtime and triggers alerts upon the following types of security events: malicious image startups, attacks by viruses or malicious programs, intrusions into containers, container escapes, and high-risk operations on containers. To use this feature, you must first activate Security Center. If you use a Resource Access Management (RAM) user, make sure that the RAM user has the permissions to access Security Center.

All regions

Use security monitoring capabilities

Scheduled backup

Scheduled backups are supported for Elastic Block Storage (EBS) devices. You can create scheduled snapshots from disks. To use this feature, you must first install the cluster-storage-operator component.

All regions

No impact on workloads

IPvlan and eBPF supported by Terway

IPvlan and extended Berkeley Packet Filter (eBPF) are supported by Terway. If an ENI is shared among pods, Terway allows you to use IPvlan and eBPF for network virtualization.

Terway enables pod network virtualization by using the lightweight IPvlan technology. This allows pod traffic to bypass the network stack of the host and reduces the network performance overheads. Terway uses Cilium as the BPF agent on nodes to configure BPF rules for pod ENIs. This enables Services and network policies to be configured on ENIs. This way, requests within pod networks are forwarded to ENIs through IPvlan. This reduces network complexity.

Note

This feature applies to the Alibaba Cloud Linux 2 operating system. To use this feature, you must submit a ticket to apply to be added to a whitelist.

All regions

Work with Terway

New regions

ACK Pro clusters are available in the China (Beijing), China (Shenzhen), Germany (Frankfurt), Indonesia (Jakarta), and China East 2 Finance regions.

China (Beijing), China (Shenzhen), Germany (Frankfurt), Indonesia (Jakarta), and China East 2 Finance

Overview of ACK Pro clusters

ACK@Edge released for commercial use

ACK@Edge is released for commercial use. ACK@Edge is a cloud-managed solution that is provided by ACK to coordinate cloud and edge computing.

All regions

ACK@Edge overview

July 2020

Feature

Description

Region

References

ACK Pro cluster released for public preview

ACK Pro clusters are released for public preview. This type of cluster is developed based on ACK managed cluster and provides higher reliability and security in large-scale production environments for enterprise users. ACK Pro clusters are also covered by SLAs that include compensation clauses. This type of cluster is suitable for the following users:

  • Internet enterprises. These enterprises deploy their business in large-scale production environments and require business management with high stability, security, and observability.

  • Big data computing enterprises. These enterprises deploy large-scale data computing services, high-performance data processing services, and other services with high elasticity. These services require clusters with high stability, high performance, and efficient computing capabilities.

  • International enterprises that run their business in China. These enterprises prioritize security and services that provide SLAs with compensation clauses.

  • Financial enterprises. These enterprises require SLAs that include compensation clauses.

All regions

Overview of ACK Pro clusters

New regions

ACK Serverless is available in the Japan (Tokyo) and Indonesia (Jakarta) regions.

Japan (Tokyo) and Indonesia (Jakarta)

ACK Serverless overview

CCM

The CCM is updated to V1.9.3.313-g748f81e-aliyun. The following features are provided:

  • Supports deletion protection for SLB instances. By default, deletion protection is enabled for newly created SLB instances.

  • Supports modification protection for the configurations of SLB instances. By default, modification protection is enabled for the configurations of newly created SLB instances.

  • Allows you to specify the resource group for an SLB instance when you create a Service.

  • Allows you to specify the name of an SLB instance when you create a Service.

  • Allows you to mount pods in Terway mode to the backend of an SLB instance.

All regions

Cloud Controller Manager

Security management

Security management is supported for your clusters. You can configure pod security policies and cluster inspections.

Pod security policy is a significant method to verify the security of pod configurations before pods are deployed. This ensures that applications are running in secure pods. Cluster inspection detects the security risks of workloads in an ACK cluster and generates inspection reports for your reference. This way, you can check whether the workloads in your ACK cluster run in a secure environment.

All regions

No impact on workloads

Shared VPCs supported by ACK

Shared VPCs are supported. A shared VPC can host cloud resources that are created by multiple accounts. The cloud resources include ECS instances, SLB instances, and ApsaraDB RDS instances. This provides a unified approach for you to manage cloud resources in a shared VPC. Shared VPCs are powered by the resource sharing mechanism. The Alibaba Cloud account that owns a shared VPC can share all vSwitches in the VPC with other accounts in the same organization. You can select a shared VPC when you create an ACK cluster. If you select a shared VPC for an ACK cluster, you can use only Terway as the network plug-in.

All regions

No impact on workloads

Cluster registration

Cluster registration is supported. During daily O&M, you may need to deploy multiple clusters in the cloud and data centers. In some scenarios, you may even deploy clusters in the clouds of different cloud service providers. In these cases, you can register external Kubernetes clusters in the ACK console. This allows you to manage external Kubernetes clusters in the console and reduce O&M costs.

All regions

Overview

Workload management

Redeployment and rollback of workloads are supported. ACK provides features on the workload management page in the ACK console, such as application redeployment and rollback. This makes it more convenient to manage your workloads.

All regions

Create a stateless application by using a Deployment

June 2020

Feature

Description

Region

References

Taint management

Taint management is supported for node pools. You can configure taints when you create or modify a node pool. This allows you to add taints to all nodes in the node pool. You can select Synchronize Node Labels and Taints to update taints for existing nodes in a node pool.

All regions

Manage node labels and taints

Application migration from virtual machines to ACK clusters

Application migration from virtual machines to ACK clusters by using Server Migration Center (SMC) is supported. SMC allows you to migrate servers to Container Registry. You can use SMC to migrate containerized applications to Container Registry at low costs.

All regions

Migrate a server to Container Registry

May 2020

Feature

Description

Region

References

Advanced security groups

Advanced security groups are supported when you create a cluster. You can select a basic security group, an advanced security group, or an existing security group. Compared with a basic security group, an advanced security group can contain up to 65,536 private IP addresses. Advanced security groups are used for clusters where a large number of containers or instances are deployed.

All regions

Create an ACK managed cluster

Component management

The Prometheus component and Kubernetes event center can be installed from the Add-ons page of the ACK console. ACK is integrated with the most commonly used Prometheus component in the container monitoring field, and the most commonly used node-problem-detector (NPD) component in the O&M field. You can select these components when you create a cluster. You can also update and maintain the components on the Add-ons page of the ACK console. The Prometheus component is provided by ARMS. NPD is a tool used for node problem detection. NPD can export events that record node exceptions, such as Docker Engine hangs, Linux kernel hangs, network access issues, and file descriptor issues. You can click the Event Center tab on the Events page to view event details.

All regions

Use Alibaba Cloud Prometheus Service to monitor an ACK cluster

Kubernetes 1.16.9

Kubernetes 1.16.9 is supported. You can create a cluster of Kubernetes 1.16.9. If the Kubernetes version of your cluster is earlier than V1.16.9, go to the Clusters page and choose More > Upgrade Cluster in the Actions column to update to Kubernetes 1.16.9. Compared with Kubernetes 1.16.6, Kubernetes 1.16.9 fixes the CVE-2020-8555 SSRF vulnerability for the kube-controller-manager component.

All regions

Vulnerability fixed: CVE-2020-8555 in kube-controller-manager

Elastic workload

Elastic workloads are supported. You can go to the App Catalog page and select ack-kubernetes-elastic-workload to install the component. You can use ACK and Virtual Kubelet in combination to proportionally schedule pay-as-you-go and preemptible instances. This allows you to schedule your workloads with elasticity.

All regions

App Marketplace

Application center

The application center feature is released in the ACK console. In earlier versions of the ACK console, after applications are deployed, the topology of the applications is not displayed in a unified view. Therefore, version management and rollback cannot be unified for continuous deployments. The application center provides a unified portal for your applications. This allows you to view the deployment of applications in a unified manner. You can also view the deployment status and changes of all ACK sub-resources that are allocated to each application. In addition, Gits and Helm charts are used to deploy applications in ACK clusters by versions. This allows you to publish or roll back different application versions deployed in ACK clusters.

All regions

Application Center overview

April 2020

Feature

Description

Region

References

AGS released for commercial use

Alibaba Cloud Genomics Service (AGS) is released for commercial use. AGS is an ACK-based big data compute service provided by Alibaba Cloud for users in the biotechnology industry. AGS provides efficient, elastic, and reliable services. AGS is faster in computing and more cost-effective than traditional methods. AGS uses the pay-as-you-go billing method and charges you based on the number of successful API calls in the backend. To submit a computing task, you only need to run a command to call the AGS API on the client. This process is counted as one API call.

All regions

AGS overview

Online expansion supported by dynamically provisioned volumes

Expansion of dynamically provisioned volumes without restarting pods is supported for Kubernetes 1.16 and later.

All regions

Expand a disk volume without service interruptions

Ingress controller

Multiple Ingress controllers can be deployed in a Kubernetes cluster. An Ingress is an important entry for Layer 7 services. If you create only one Ingress for a cluster, the routing performance may encounter a bottleneck. If an Ingress allows inbound access through the Internet and private network at the same time, security risks exist. To solve these issues, ACK provides a Helm chart for the Ingress controller when only one Ingress is used. The name of the Helm chart is ack-ingress-nginx. You can deploy multiple Ingress controllers from App Catalog. You can use YAML files to configure access to Internet-facing and internal-facing SLB instances separately.

All regions

Deploy a high-reliability NGINX Ingress controller

New region

ACK Serverless is available in the India (Mumbai) region.

India

Create an ACK Serverless cluster

March 2020

Feature

Description

Region

References

Component management

The following features are added for component management:

  • Allows you to view the YAML files of components.

  • Allows you to perform health checks for nodes before component updates. This prevents component update failures that are caused by node drains or exceptions.

  • Allows you to manually refresh the Add-ons page.

All regions

Manage system components

Self-managed ECS instances

Nodes that run on self-managed ECS instances can be added to the backend of SLB instances by using the CCM. This way, the existing applications and containerized applications share the same SLB instances and inbound traffic. This is suitable for scenarios where existing applications are gradually replaced by containerized applications.

All regions

Cloud Controller Manager

Cluster expansion and node specification changes

Cluster expansion and node specification changes are supported by Terway. When you manually expand a cluster, you may need to create nodes in new zones. In earlier versions, to create pods in a new zone, you must first add new pod vSwitches in the zone. You can add pod vSwitches in Terway ConfigMaps. When you change the specifications of a node, the maximum number of pods that are supported by Terway on the node also changes. After this release, the K8s max-pod parameter is automatically adjusted to fit the new node specifications.

All regions

Work with Terway

Node pool management

Node pool management is supported. A node pool contains a group of nodes with the same configurations. For example, nodes in a node pool are configured with the same container runtime, OS, and security group. You can create multiple node pools for a cluster. This allows you to deploy a variety of services to different node pools in a cluster. Node pools also support auto scaling. Nodes can be automatically added when a node pool is short of required resources.

All regions

Create and manage a node pool

Cluster inspection

Cluster inspection is optimized. Cluster inspection is the core feature provided by ACK for cluster O&M. Cluster inspection dynamically scans clusters to identify potential risks. The optimization provides the following features:

  • Displays information about unknown hosts.

  • Checks the availability of Yellow dogUpdater, Modified (YUM).

  • Checks the availability of systemd.

All regions

Work with cluster check

Kubernetes 1.16

Update to Kubernetes 1.16.6 is supported. You can update your clusters from Kubernetes 1.14.8 to 1.16.6. You can also create clusters that run Kubernetes 1.16.6. We recommend that you read the update notes before you update your clusters.

All regions

Update the Kubernetes version of an ACK cluster

New region

ACK managed clusters are available in the China South 1 Finance region.

China South 1 Finance

Create an ACK managed cluster

ephemeral-storage

The ephemeral-storage parameter is added for container configurations when you create an application. Ephemeral storage is a new storage resource similar to CPU and memory resources. Kubernetes uses this parameter to manage and schedule the transient storage of applications that run in Kubernetes clusters. The root directory and log directories (/var/log) of kubelet are stored on the primary partition of a node. In addition, emptyDir volumes, container log, image layers, and the writable layers of containers are also stored on the primary partition. Therefore, ephemeral-storage is used to manage the primary partition of a node. You can set requests and limits when you create an application. This allows you to schedule and manage the storage resources that are allocated from the primary partition to the application.

All regions

Create a stateless application by using a Deployment

February 2020

Feature

Description

Region

References

Kubernetes 1.16 and Docker 19.03.5

Kubernetes 1.16 and Docker 19.03.5 are supported to provide enhanced cloud-native capabilities. Compared with the earlier version, Kubernetes 1.16 accelerates pod creation and improves affinity, stability, and observability. You can select Docker 19.03.5 when you create a cluster. ACK accelerates container startups and the building of images that are based on Docker 19.03.5.

All regions

(Discontinued) Kubernetes 1.16 release notes

Auto scaling

The auto scaling feature is updated to add the following options: the Alibaba Cloud Linux2 operating system, custom security groups, and preemptible instances with GPU capabilities. To use Alibaba Cloud Linux 2 and custom security groups, you must first submit a ticket to enable them for your account.

All regions

submit a ticket

CentOS 7.7

CentOS 7.7 is supported as the node OS. You can specify the CentOS 7.7 operating system when you create worker nodes. CentOS 7.7 is automatically used when you expand clusters or enable auto scaling for clusters.

All regions

submit a ticket

Helm 3

Helm 3 is supported. You can install Helm 3 from App Catalog. Compared with Helm 2, Helm 3 improves the security of role assignment, provides full compatibility with Kubernetes role-based access control (RBAC) in multi-tenant scenarios, and supports hooks for more management operations.

All regions

For more information about how to update from Helm 2, see [Component Updates] Update Helm V2 to V3.

New regions

ACK Serverless is available in the Indonesia (Jakarta) and UK (London) regions. You can create ACK Serverless clusters in these regions in the ACK console.

Indonesia (Jakarta) and UK (London)

Create an ACK Serverless cluster

ClusterIP Services in ACK Serverless clusters

ClusterIP Services are supported in ACK Serverless clusters. This provides more options when you deploy containerized applications in ACK Serverless clusters. You can create ClusterIP Services in an ACK Serverless cluster to enable access to your workloads from within the ACK Serverless cluster.

All regions

Use Services to expose applications

CCM

ECS instances and elastic container instances can be attached to the backend of SLB instances that are associated with Services by using the CCM. This enables unified scheduling for application pods across worker nodes and virtual nodes. This also improves application resilience.

All regions

Release notes for the CCM

ACK edge clusters

32-bit and 64-bit ARM nodes are supported in ACK edge clusters. This allows ACK edge clusters to support more heterogeneous infrastructures. You can add Edge Node Service (ENS) nodes or nodes from data centers to ACK edge.

All regions

Add an edge node

January 2020

Feature

Description

Region

References

Virtual nodes

ClusterIP Services can be accessed by pods that are deployed on virtual nodes. This enables Kubernetes to centrally manage virtual nodes and elastic container instances. You can deploy applications on virtual nodes without the inconvenience of resource capacity planning. This meets the requirements of scenarios such as online workload scaling, offline computing, and CI/CD, and also reduces the overall computing costs. To enable this feature, log on to the console, click App Catalog, and then find and install ack-virtual-node.

All regions

Schedule pods to elastic container instances through virtual nodes

API server

Service account token volume projection can be enabled for the API server when you create a cluster. This enables service account authentication on pods. This feature is also required if mutual Transport Layer Security (TLS) authentication is enabled on Istio through Secret Discovery Service (SDS).

All regions

Create an ACK dedicated cluster (discontinued)

CSI

CSI can be selected as the volume plug-in when you create an ACK cluster. The optimized CSI plug-in provides the following features:

  • OSS subdirectories can be mounted to containers.

  • The Memory type emptyDir volumes are supported. The Memory type volume is a RAM-based temporary file system, whose storage space is limited by memory. This type of file system provides good performance and is typically used to provide caching space in containers.

  • Accelerated OSSFS transmission is supported. OSSFS allows you to share data by mounting OSS buckets to local file systems in Linux. To meet the requirements of big data and AI scenarios, ACK improves read speed by adjusting concurrency, block size, and libfuse configurations. For more information, see alibaba-cloud-csi-driver.

All regions

Install CSI

Sandboxed containers

Disks and NAS file systems can be mounted to sandboxed containers to enhance cloud-native capabilities. This allows ACK to provide the same storage performance as when these storage services are used on virtual machines. ACK also supports RootFS BLKIO Limit and disk I/O throttling on pods, and optimizes its support for multi-tenancy.

All regions

Mount a NAS file system to a sandboxed container

ACK clusters for confidential computing

ACK clusters for confidential computing are released for public preview. This type of cluster is developed on top of Intel Software Guard Extensions (SGX) and is particularly suitable for sensitive data protection and scenarios such as smart contracts in blockchains, user secrets processing, intellectual property protection, genomics computing in bioinformatics, and edge computing. You can create and manually expand ACK clusters for confidential computing. You can also enable auto scaling, and add different types of nodes to the clusters. For more information, see Create an ACK managed cluster for confidential computing and SGX application development guide. ACK also provides open source sgx-device-plugin to help you deploy SGX applications in ACK clusters. For more information, see Kubernetes device plugin for Intel SGX.

Note

Intel (R) SGX is a set of CPU instruction codes that are developed by Intel. Intel (R) SGX allows you to run application code and data in a special runtime environment called enclave, which is built on top of hardware silos and memory encryption technologies. Enclaves refer to Trusted Execution Environment (TEE). No application, OS Kernel, BIOS, or hardware other than the CPU can access an enclave without verification. All regions data in the enclave memory is encrypted. Users encrypt the code and data in an enclave with their private keys obtained from Intel. An enclave can be started only after the signature is verified through Intel Attestation Service (IAS), which is a remote certification service of Intel.

All regions

Create an ACK managed cluster for confidential computing

AGS

Gene sequencing is supported by calling AGS API operations. ACK has released a set of AGS API operations. You can call these API operations to submit gene sequencing tasks. Results are automatically uploaded to your OSS buckets. This saves you the inconvenience of cluster creation and task deployments. These API operations support different SLA levels and provide computing resources based on different requirements. This allows you to reduce costs and improve efficiency. This feature is in public preview. To use the feature, submit a ticket.

All regions

Use AGS to process WGS tasks

December 2019

Feature

Description

Region

References

Component management

Component management is supported. You can log on to the ACK console. On the Clusters page, find the cluster that you want to manage and choose More > Manage System Components in the Actions column to manage cluster components. You can manage all system components and optional components by performing operations such as update, uninstall, and reinstall. Custom component configurations will be available soon.

All regions

Manage system components

App Catalog

The ack-node-local-dns plug-in is provided in App Catalog to speed up Domain Name Service (DNS) queries. ack-node-local-dns sends internal DNS queries to CoreDNS and directly forwards external DNS queries to external DNS resolvers. ack-node-local-dns caches all queries and provides DNS caching on each node. This significantly improves the overall DNS query rate of the cluster.

All regions

App Marketplace

New region

ACK managed clusters are available in the China East 1 Finance region. You only need to create worker nodes in an ACK managed cluster. ACK creates and manages master nodes. This type of cluster is easy to use and provides high availability at low costs. This saves you the inconvenience of master node O&M and allows you to focus on business development.

China East 1 Finance

Create an ACK managed cluster

NPU-accelerated ECS instances

Neural processing unit (NPU)-accelerated ECS instances are supported when you create ACK managed clusters or ACK dedicated clusters. The instance type is ecs.ebman1.26xlarge, which is suitable for big data analytics and AI scenarios in video and graphics industries.

All regions

Create an ACK managed cluster

Terway

The user experience of Terway is improved. The new user interface provides information about the number of pods that are supported by each ECS instance type when you create a cluster. When you expand a cluster, the user interface also provides multiple options. This allows you to select vSwitches for nodes and pods. The user interface is optimized to provide easy-to-read, accurate information.

All regions

Work with Terway

November 2019

Feature

Description

Region

References

Cluster expansion

Multiple zones and multiple data disks are supported when you expand an ACK cluster. The user interface for expanding an ACK cluster is updated to provide the same configuration options as those for creating an ACK cluster. You can select multiple zones when you expand an ACK cluster. You can also mount multiple data disks to a node and specify whether to encrypt these disks.

All regions

Increase the number of nodes in an ACK cluster

Custom node configurations

Custom scripts, tags, and Operation Orchestration Service (OOS) are supported for node configurations. You can write custom scripts to configure nodes when you create or expand an ACK cluster. To use this feature, submit a ticket to enable this feature for your account. You can use this feature to specify the node OS. Instead of building custom images, you can directly inject scripts into standard images. Auto scaling allows you to add tags to cluster nodes. This makes it easier for you to identify cluster nodes and allocate the cost of nodes. ACK integrates OOS into the node O&M. You can go to the OOS page from the ACK console and run OOS scripts to maintain nodes on the OOS page.

All regions

Increase the number of nodes in an ACK cluster

Multiple zones and log auditing supported in ACK Serverless clusters

Multiple zones and log auditing are supported in ACK Serverless clusters. After ACK Serverless is updated to V2.0, ACK Serverless clusters provide more cloud-native features. Cross-zone ACK Serverless clusters and log auditing are supported. You can deploy pods across zones to improve the availability of your business. You can also use log auditing to improve the security of ACK Serverless clusters. ACK Serverless clusters will be improved to provide the same features as dedicated and ACK managed clusters.

All regions

Create an ACK Serverless cluster

vGPU

vGPU resources are provided through the vgn5i instance family to meet the requirements of AI and big data industries. You can select instance types of the vgn5i instance family when you create an ACK cluster.

All regions

No impact on workloads

ENI buffer pools for Terway

ENI buffer pools are supported for Terway. Terway is a container network plug-in that is developed on top of Alibaba Cloud ENI. The update enables Terway to create a buffer pool of ENI IP addresses during node initialization. This accelerates pod creation and improves user experience.

All regions

Work with Terway

CCM

External ECS instances can be added to the backend of SLB instances by using the CCM. The CCM is a system component that associates Services with SLB instances. By default, cluster nodes that host Services are mounted to the backend of the related SLB instances. The update allows you to add ECS instances outside an ACK cluster as the backend servers to the related SLB instances. This makes it easier to perform application migration and canary releases.

All regions

Cloud Controller Manager

October 2019

Feature

Description

Region

References

Alibaba Cloud Linux2

The Alibaba Cloud Linux2 operating system is supported. Alibaba Cloud Linux2 is the latest OS version that is developed by Alibaba Cloud on top of an advanced CentOS kernel version. Alibaba Cloud Linux2.1903 is fully adapted to ACK. This OS version supports faster startups and optimized performance, and improves the efficiency and reliability of ACK clusters.

All regions

Create an ACK dedicated cluster (discontinued)

Ingress dashboard

The Ingress dashboard is provided. In earlier versions, you must manually configure the Ingress dashboard, which is a time-consuming and error-prone task. A check box is added to the configuration page of the Ingress controller. You need to select the check box to enable the Ingress dashboard feature. This way, the Ingress dashboard is automatically installed after the cluster is created.

All regions

Create an ACK dedicated cluster (discontinued)

SLB instance specifications

Multiple SLB instance specifications are supported when you create a Service. In earlier versions, when you create a LoadBalancer Service, ACK automatically creates shared-performance SLB instances. To meet your requirements in various scenarios, ACK allows you to select SLB instance specifications when you create a LoadBalancer Service. The SLB instances adopt the pay-as-you-go billing method.

All regions

Use Services to expose applications

EIPs for the API server

An EIP can be associated to or disassociated from the API server of an ACK cluster. SLB instances provide access to the API server of an ACK cluster. When you create an ACK cluster, ACK allows you to specify an Internet-facing or internal-facing SLB instance to handle traffic to the cluster. However, you may need to change the network type of the SLB instance after the cluster is created. ACK allows you to associate an EIP with or disassociate an EIP from the SLB instance after the cluster is created. This allows you to change the access mode to the API server between Internet access and internal access.

All regions

Create an ACK dedicated cluster (discontinued)

Auto scaling of ENS nodes for ACK edge clusters

The auto scaling of ENS nodes in ACK edge clusters is supported. To support edge computing scenarios, ACK allows you to configure auto scaling of ENS nodes in ACK edge clusters. This feature can be implemented by calling the API.

All regions

Enable node auto scaling

New region

ACK Serverless is available in the China (Zhangjiakou) region.

China (Zhangjiakou)

Create an ACK Serverless cluster

September 2019

Feature

Description

Region

References

New region

ACK is available in the China (Chengdu) region. You can create ACK dedicated clusters in the China (Chengdu) region.

To create ACK managed clusters in the China (Chengdu) region, submit a ticket.

China (Chengdu)

Create an ACK dedicated cluster (discontinued)

Kubernetes 1.14.6 and new features for cluster updates

The canary release of the update from Kubernetes 1.14.6 is implemented in the following regions: China (Shanghai), China (Zhangjiakou), Singapore, and Germany (Frankfurt). Updates from Kubernetes 1.14.6 will soon be available in all regions. More features are also provided to simplify the update process. In the ACK console, you can click Upgrade Cluster on the Clusters page to update your cluster.

The new update feature adds the following improvements to secure updates:

  • A comprehensive cluster check is performed before an update.

  • You can manually pause or resume an update.

  • Detailed log of updates is retained.

  • China East 2 Finance

  • China (Zhangjiakou)

  • Singapore

  • Germany (Frankfurt)

Update the Kubernetes version of an ACK cluster

Node maintenance

Node maintenance is supported. To maintain nodes in a cluster, you must make sure that workloads are not deployed on the nodes that you want to maintain. ACK supports node maintenance.

You can select one or more nodes that you want to maintain and set them to unschedulable on the Nodes page. You can also drain these nodes.

  • After you set a node to unschedulable, pods cannot be scheduled to the node.

  • If you drain a node, no new pods are scheduled to the nodes and existing pods on the node are migrated to other nodes. However, pods that are managed by DaemonSets are not migrated from the node.

If you have a LoadBalancer Service, you can specify whether to remove nodes that run the pods that are associated with the Service from the backend of the related SLB instance when these nodes are set to unschedulable. This allows you to flexibly manage your workloads during node maintenance.

All regions

Set node schedulability

Custom node names

Custom node names are supported. To manage a cluster that includes a large number of nodes, you must identify nodes by name. The default node names provided by ACK are not easy to identify. ACK allows you to customize node names when you create a cluster. When you create a cluster in the ACK console, you can select Custom Node Name in the advanced settings of the cluster. You can define a prefix, an IP substring length, and a suffix for a custom node name. The IP substring length specifies the number of digits to be truncated from the end of a node IP address and can be used to uniquely identify a node.

All regions

Create an ACK dedicated cluster (discontinued)

Advanced security groups

Advanced security groups are supported when you create an ACK clusters. Compared with basic security groups, advanced security groups support more ECS instances, more ENIs, and effective management on an infinite number of private IP addresses. Advanced security groups are suitable in scenarios that require high O&M efficiency, high ECS instance specifications, and a large number of compute nodes. To meet the requirements of a large-scale cluster, you can select advanced security groups for Security Group in the advanced settings when you create the cluster.

All regions

Create an ACK dedicated cluster (discontinued)

Disk encryption and CSI

Disk encryption and the CSI component are supported. ACK allows you to encrypt data disks. You can enable disk encryption for the selected data disks when you create a cluster. This feature can automatically encrypt the data that is transmitted from an ECS instance to a data disk and automatically decrypt the data when it is read. This improves data security. In addition, Kubernetes 1.14.6 supports the standard CSI plug-in, which is generally used for volume management. You can select FlexVolume or CSI when you create a cluster.

All regions

Create an ACK dedicated cluster (discontinued) and Storage

August 2019

Feature

Description

Region

References

Kubernetes 1.14.6

Kubernetes 1.14.6 is supported. You can select Kubernetes 1.14.6 when you create a cluster in the ACK console. You cannot update an existing cluster to Kubernetes 1.14.6.

All regions

No impact on workloads

New regions

ACK Serverless is available in the Singapore, China (Hong Kong), and Australia (Sydney) regions. ACK Serverless allows you to create containerized applications without managing or maintaining clusters and nodes. You are billed based on the actual amount of resources that are consumed by the elastic container instances that run the applications.

ACK Serverless clusters allow you to focus on the design and development of applications, instead of managing the underlying infrastructures.

Singapore

China (Hong Kong)

Australia (Sydney)

Create an ACK Serverless cluster

ACK Serverless 2.0

ACK Serverless 2.0 is released to provide more Kubernetes-native features. ACK Serverless 2.0 supports multiple namespaces, CRD objects, RBAC, PVs, and PVCs. ACK Serverless 2.0 improves the security and isolation capability of clusters. The average price of ACK Serverless clusters is reduced by 46% due to lower costs of elastic container instances. This includes a 30% reduction in CPUs and a 65% reduction in memory.

All regions

Create an ACK Serverless cluster

SCC

ACK clusters based on Super Computing Cluster (SCC) resources are supported. SCCs are powered by ECS Bare Metal (EBM) instances and use the high-speed Remote Direct Memory Access (RDMA) technology. SCCs improve network performance. SCCs are used in scenarios such as high-performance computing, AI, machine learning, scientific and engineering computing, data analytics, and audio and video processing. You can create SCC-based ACK clusters. This type of cluster combines high-performance infrastructure resources with lightweight and agile containers. SCC-based ACK clusters are applicable to high network throughput and compute-intensive scenarios.

All regions

Create an ACK dedicated cluster (discontinued)

Auto scaling and cross-zone scheduling

Multiple scaling groups are supported for auto scaling. Cross-zone scheduling policies are supported. The auto scaling feature is optimized. You can configure multiple scaling groups so that resources of different specifications are automatically added when the scaling threshold is reached. This feature meets the requirements of running compute-intensive applications and GPU computing tasks. When you configure auto scaling policies, you can specify different scheduling policies for multiple zones, including priority policies, cost optimization policies, and zone balancing policies. This meets the requirement for resource scheduling when the cluster is deployed across multiple zones.

All regions

Enable node auto scaling

Custom cluster domain names

Custom cluster domain names are supported. ACK allows you to customize a cluster domain name by specifying the cluster-domain parameter. The cluster-domain parameter specifies the local domain name that is used for service discovery. If you have multiple clusters, we recommend that you customize the local domain names to simplify the management of clusters and services. ACK allows you to customize a cluster domain name when you create a cluster. This simplifies management and improves the O&M efficiency.

All regions

Create an ACK dedicated cluster (discontinued)

App Hub

App Hub is provided in App Catalog. App Hub provides various cloud-native and open source containerized applications. ACK integrates App Hub into App Catalog. To deploy cloud-native applications in your cluster, log on to the ACK console and click the App Hub tab on the App Catalog page to find and install the applications with one click. This saves you the inconvenience of creating clusters and deploying applications by using a CLI.

All regions

App Marketplace

CCM

The CCM is updated. The CCM is the core component in an ACK cluster and is responsible for managing various cloud resources, such as SLB instances and VPCs. The following features are added to the CCM:

  • SLB instances can be created with access control settings. You can specify an IP whitelist for an SLB instance that is created by ACK. This enhances the security of the ACK cluster.

  • You can specify whether to remove unschedulable nodes when you run the kubectl cordon or kubectl drain command. Cordoning and draining nodes are important features in cluster maintenance. However, the community has not reached an agreement on whether to remove a node from the backend of an SLB instance when the node is set to unschedulable for maintenance. The CCM provides an interface that allows you to specify whether to remove such nodes from the backend of the SLB instance. This ensures the flexibility of maintenance.

  • Pods can be mounted to the backend of an SLB instance by using Terway. Terway ENI is the latest network plug-in that is provided by ACK. The core feature of Terway ENI is to mount the ENI IP address of a node to a pod. The CCM allows you to mount pods instead of nodes to the backend of an SLB instance. This prevents traffic forwarding through nodes and improves network performance.

  • Node weights can be set based on the number of pods on each node for Services in Local mode. The CCM can adjust the percentage of traffic that is sent to each node based on the number of pods on each node. This balances workloads among nodes. This feature applies to only Services in Local mode.

All regions

Cloud Controller Manager

July 2019

Feature

Description

Region

References

ACK edge clusters

ACK edge clusters are released for public preview. You can add edge nodes or ENS nodes to ACK edge clusters. This type of cluster supports edge computing and manages edge nodes and ENS nodes to reduce O&M costs. This type of cluster also supports autonomous edges and networks to meet the requirements in different edge computing scenarios. You can select this type of cluster on the cluster template page.

China site

-

Multi-cluster management

The multi-cluster management feature is released for public preview. You can select Register Kubernetes Cluster on the cluster template page to add Kubernetes clusters from data centers and other public clouds to the ACK console. Then, you can deploy applications to these clusters in the console. You can manage hybrid cloud clusters and clusters that are deployed across multiple clouds. After you add self-managed clusters from data centers to ACK, you can manage these clusters by using the O&M feature that is provided by ACK.

China site

Register an external Kubernetes cluster

New region

ACK managed clusters are available on the Alibaba Cloud Japan site.

  • Saves resources.

    You do not need to create master nodes in an ACK managed cluster. If you use another type of cluster, you must create at least three master nodes.

  • Improves O&M efficiency.

    ACK manages master nodes.

  • Ensures security.

    ACK meets various security requirements.

Japan

Create an ACK managed cluster

Support for multiple data disks during cluster creation

Multiple data disks can be mounted to nodes when you create an ACK cluster. This saves you the inconvenience of manually adding data disks after the cluster is created. ACK formats and mounts one of the selected data disks to the docker directory. You can determine how to handle the other data disks.

All regions

Create an ACK dedicated cluster (discontinued)

Existing security groups selectable during cluster creation

An existing security group can be selected when you create an ACK cluster. You can specify an existing security group for the VPC of your cluster in the advanced settings. This allows you to use custom inbound and outbound security group rules to improve the security of your cluster.

All regions

Create an ACK dedicated cluster (discontinued)

Deletion protection

Deletion protection is released to ensure the security of your cluster. You are required to enter a Short Message Service (SMS) verification code when you delete a cluster. However, you may mistakenly delete the cluster by calling the API. To ensure the security of clusters, ACK supports deletion protection for clusters. You can enable deletion protection when you create a cluster. This way, you cannot delete the cluster in the console or by calling the API. To delete the cluster, you must first disable deletion protection. You can enable or disable deletion protection on the cluster details page.

All regions

Create an ACK dedicated cluster (discontinued)

Batch authorization

Multiple RAM users can be authorized at the same time. You can also grant the permissions to manage all clusters. This allows you to efficiently authorize RAM users. The authorization procedure is also optimized to improve user experience.

All regions

Authorization

Time zone

The time zone of an application can be synchronized to that of the node. You can select Synchronize Timezone from Node to Container when you create an application from an image. This ensures that the application pods and the host node use the same time zone.

All regions

Create a stateless application by using a Deployment

New region

Container Registry Enterprise Edition is available in the UK (London) region. Container Registry Enterprise Edition supports large-scale image distribution with enhanced security. This service is suitable for enterprise users that require high security and large-scale nodes.

UK (London)

What is Container Registry?

Helm V2 charts supported by Container Registry Enterprise Edition

Helm V2 charts are supported by Container Registry Enterprise Edition to make it easier for you to manage cloud-native assets. You can enable the charts component on the Overview page of your Container Registry Enterprise Edition instance. When the component is running, you can start to manage Helm chart repositories.

All regions

None

June 2019

Feature

Description

Region

References

New regions

ACK managed clusters are available in the Japan (Tokyo) and UK (London) regions on Alibaba Cloud public cloud.

Japan (Tokyo)

UK (London)

What is ACK?

Terway

A new version of Terway is released. The exclusive ENI mode and the inclusive ENI mode are supported by this version. The default mode is the inclusive ENI mode.

  • The exclusive ENI mode: In this mode, the number of pods that can be deployed on a node must match the number of ENIs that can be created on the node. This mode improves network performance.

  • The inclusive ENI mode: In this mode, you can deploy multiple pods on a node. The pods share the same ENI.

All regions

Work with Terway

Knative

Knative is supported. Knative is a Kubernetes-based serverless framework. Knative creates a cloud-native and cross-platform orchestration standard for serverless applications. Knative implements this standard by integrating the creation of containers (or functions), workload management (auto scaling), and event models. ACK supports Knative and allows you to install and update the Build, Serving, and Eventing components. You must deploy Istio before you use Knative. ACK provides instructions to deploy sample applications, and also provides best practices for tracing, monitoring, and logging applications.

All regions

Knative and Quickly deploy a Knative application

Pod search

Pods can be searched by node IP address or pod IP address. In the ACK console, choose Applications > Pods and specify a node IP address or a pod IP address to search for a pod. This saves the time to find pods that you want to manage and maintain.

All regions

No impact on workloads

May 2019

Feature

Description

Region

References

New regions

ACK managed clusters are available in the Australia (Sydney) region on Alibaba Cloud public cloud and the China East 2 Finance region on Alibaba Finance Cloud.

You can create ACK managed clusters in the Australia (Sydney) region on Alibaba Cloud public cloud and the China East 2 Finance region on Alibaba Finance Cloud.

Australia (Sydney)

China East 2 Finance

What is ACK?

Genomics computing clusters for genomics computing

Genomics computing clusters are released. This type of cluster uses high-performance computing (HPC) instances as worker nodes and provides a large-scale workflow engine for batch genomics computing. Genomics computing clusters are suitable for data splitting and mutation detection, and support data analytics for the following formats: BCL, FASTQ, BAM, SAM, and VCF. In the ACK console, choose Clusters > Clusters and click Create Kubernetes Cluster. In the Select Cluster Template dialog box, select Genomics Computing Cluster.

All regions

No impact on workloads

ACK clusters with FPGA-accelerated nodes

ACK clusters with FPGA-accelerated nodes are released. This type of cluster uses FPGA F3 instances as worker nodes and is used for H265 video encoding and image conversion from JPEG to HEIF. FPGA-based video encoding reduces the processing time from more than 1 week to 15 minutes. This significantly reduces the bitrate and saves bandwidth costs when transcoding videos of the same quality. In the ACK console, choose Clusters > Clusters and click Create Kubernetes Cluster. In the Select Cluster Template dialog box, select Dedicated FPGA Cluster to create an ACK dedicated cluster with FPGA-accelerated nodes.

All regions

No impact on workloads

CCM

The CCM is updated to V1.9.3.110-g4938309-aliyun. This version supports more SLB configuration options. The following features are provided:

  • Allows you to restrict the creation of Internet-facing SLB instances by setting parameters.

  • Allows you to change certificate IDs.

  • Allows you to specify a vSwitch when you attach an internal-facing SLB instance to a Service.

  • Allows you to set SLB instance configurations to redirect traffic from HTTP port 80 to HTTPS port 443.

All regions

Cloud Controller Manager

Istio

Istio is updated to V1.1.4. Istio 1.1.4 improves self-recovery capabilities, and supports automatic recovery of the control plane and automatic updates of earlier versions. Istio is also integrated with Time Series Database (TSDB). TSDB is a database service that supports high-speed read and write operations, compressed storage, and real-time computing. To fix the local storage issues in Prometheus, TSDB provides remote storage services with high performance and high reliability at low costs.

Compared with other remote storage solutions provided by the community, TSDB is easier to use and only requires you to change the Prometheus configuration. The solution supports parallel read and write operations and is highly compatible with PromQL. TSDB is a distributed storage system with auto scaling capabilities.

All regions

No impact on workloads

Container Registry Enterprise Edition

Images can be synchronized across all regions worldwide for instances of Container Registry Enterprise Edition. This solves issues in the global delivery of applications and improves the business iteration efficiency for enterprises. Container Registry Enterprise Edition supports large-scale image distribution with enhanced security. It is suitable for enterprises that require high security and a large number of nodes.

All regions

No impact on workloads

Support for multiple zones and five master nodes during cluster creation

Multiple zones and five master nodes are supported when you create an ACK dedicated cluster. This allows you to create a cross-zone ACK dedicated cluster with five master nodes to significantly improve the availability of the cluster.

All regions

No impact on workloads

April 2019

Feature

Description

Region

References

Kubernetes 1.12.6

ACK managed clusters or ACK dedicated clusters in all regions can be updated from Kubernetes 1.11.5 to 1.12.6 in the ACK console.

All regions

No impact on workloads

Audit logs

Audit logs can be collected from ACK managed clusters. An audit log records operations on the API server and allows cluster administrators to trace the activities of different users.

All regions

Work with cluster auditing

Istio

Istio is updated to V1.1. Istio 1.1 allows you to manage Istio applications in the ACK console. You can create and manage Istio applications and services on a graphical interface. You can create different application versions, implement canary releases, set canary release policies, and also configure fault injection policies.

All regions

No impact on workloads

GPU-accelerated pods supported by ACK Serverless

GPU-accelerated pods are supported when you create applications in an ACK Serverless cluster. When you create an application from a template, specify the pod type as GPU in the YAML file.

All regions

No impact on workloads

Container Registry Enterprise Edition

Container Registry Enterprise Edition is available in the China (Beijing) region.

China (Beijing)

What is Container Registry?

ACK clusters with FPGA-accelerated nodes

ACK clusters with FPGA-accelerated nodes are released. This type of cluster uses FPGA F3 instances as worker nodes and is used for H265 video encoding and image conversion from JPEG to HEIF. FPGA-based video encoding reduces the processing time from more than 1 week to a short period of time. This significantly reduces the bitrate and reduces bandwidth costs when transcoding videos of the same quality. In the ACK console, choose Clusters > Clusters and click Create Kubernetes Cluster. In the Select Cluster Template dialog box, select Dedicated FPGA Cluster to create an ACK dedicated cluster with FPGA-accelerated nodes.

All regions

No impact on workloads

March 2019

Feature

Description

Region

References

New regions

ACK managed clusters are available in the China (Zhangjiakou), China (Hohhot), US (Silicon Valley), and Germany (Frankfurt) regions.

China (Zhangjiakou)

China (Hohhot)

Germany (Frankfurt)

US (Silicon Valley)

What is ACK?

Container Registry Enterprise Edition

Container Registry Enterprise Edition was officially released at the Alibaba Cloud Summit on March 21, 2019. This edition provides higher security and supports large-scale image distribution. Container Registry Enterprise Edition is in public preview in the China (Shanghai) region. To use this edition, submit a ticket.

China East 2 Finance

What is Container Registry?

Container Registry Shared Edition

Container Registry Shared Edition is available in all regions on the International site (alibabacloud.com).

All regions

What is Container Registry?

Kubernetes 1.12.6

Kubernetes 1.12.6 is supported. You can create a cluster of Kubernetes 1.12 in the console.

All regions

Create an ACK dedicated cluster (discontinued)

Log Service

The Log Service plug-in is supported by ACK managed clusters. You can enable Log Service when you create an ACK managed cluster or an ACK dedicated cluster. After the plug-in is installed, you can use Log Service to manage Kubernetes log.

All regions

Create an ACK managed cluster

New region

ACK managed clusters that run Windows are available. You can create this type of cluster in the ACK console or by calling the API. This way, you can create Windows containers and deploy traditional Windows applications on cloud-native platforms to achieve agility and elasticity.

All regions

Windows clusters are no longer supported.

IPVS

The IP Virtual Server (IPVS) proxy mode is supported. Compared with the traditional iptables mode, the IPVS mode significantly improves the load balancing performance in large-scale clusters. You can use this mode in all clusters and all regions.

All regions

Create an ACK dedicated cluster (discontinued)

Cluster templates

Multiple cluster templates are provided in the console. You can select templates of different cluster types based on your business requirements. Templates of the following cluster types are supported: ACK managed clusters, clusters with EBM instances, GPU-accelerated clusters, and Windows clusters. Cluster templates allow you to create ACK clusters based on your business requirements.

All regions

No impact on workloads

Elastic Container Instance

High-specification elastic container instances are provided for genomics computing. The maximum CPU specification is increased from 8 vCPUs to 64 vCPUs. The highest specification of an elastic container instance is 64 vCPUs and 256 GiB memory. The lowest specification of an elastic container instance is 0.25 vCPU and 0.5 GiB memory. You can select a specification based on your business requirements to achieve the highest cost efficiency.

All regions

Limits

February 2019

Feature

Description

Region

References

New region

ACK managed clusters are available in the China (Shenzhen) region. ACK managed clusters provide the following core benefits:

  • Saves resources. You do not need to create master nodes in an ACK managed cluster. Compared with other cluster types, this cluster type saves you the costs of three master nodes.

  • Improves O&M efficiency. ACK manages the master nodes.

  • Ensures security. ACK meets various security requirements.

China South 1 Finance

Create an ACK managed cluster

App Catalog

Knative add-ons are provided in App Catalog. Knative is a scale-to-zero and request-driven computing runtime based on Kubernetes and Istio. Knative supports the deployment of serverless applications and functions.

ACK provides Knative add-ons to help you build the Knative Serving environment in your cluster.

All regions

Knative

Cluster inspection

Cluster inspection is supported. You can use this feature to perform in-depth checks on cluster resources, components, and configurations. This can identify the causes of errors in your cluster.

Regions in the Chinese mainland

Work with cluster check

January 2019

Feature

Description

Region

References

Windows containers

Windows containers are supported. This allows you to deploy and run Windows applications in containers of ACK clusters. This enables Kubernetes-based elastic scheduling and management of Window applications.

You can add Windows nodes to managed and ACK dedicated clusters.

Container Registry Enterprise Edition is in private preview. To use this service, submit a ticket.

All regions

Create a Windows node pool

Container Registry Enterprise Edition

Container Registry Enterprise Edition is released for internal preview. Container Registry Enterprise Edition provides container image repositories built on top of dedicated resources. This edition provides stable image building, large-scale image distribution, and image hosting with enterprise-class security. It is suitable for enterprises that require high security and a large number of nodes.

Container Registry Enterprise Edition is in private preview. To use this service, submit a ticket.

All regions

What is Container Registry?

Intelligent cluster O&M

Intelligent cluster O&M is available in the China (Hangzhou) region. Intelligent O&M provides the best practices for cluster management in different scenarios. This allows you to identify the causes of errors in the cluster by performing in-depth checks on cluster resources, components, and configurations.

China (Hangzhou)

Work with cluster check

ARMS

ARMS is supported and integrated into ACK. After you install the ARMS plug-in, you can monitor the application performance in your cluster.

ARMS is a monitoring service for application performance management (APM). To monitor a Java application, you need only to attach an ARMS agent to the startup script of the application. No code change is required. ARMS enables you to locate failed API operations or slow calls, reproduce API parameters, detect memory leaks, and discover system bottlenecks. This significantly improves the efficiency of service diagnostics.

All regions

Monitor application performance

Elastic Container Instance

Starting January 22, 2019, you are charged for the commercial use of Elastic Container Instance. Elastic container instances are deployed as the underlying infrastructures of ACK Serverless cluster. You are charged when you create elastic container instances in ACK Serverless clusters. ACK Serverless clusters remain free of charge.

All regions

Billing description

New regions

ACK Serverless clusters are available in the China (Beijing) and China (Shenzhen) regions. ACK Serverless clusters provide improved experience with serverless containers.

China (Beijing)

China South 1 Finance

Create an ACK Serverless cluster

December 2018

Feature

Description

Region

References

New region

ACK is available in the UK (London) region on both the China site (aliyun.com) and the International site (alibabacloud.com).

UK (London)

Create an ACK dedicated cluster (discontinued)

New region

ACK managed clusters are available in the China (Shanghai), Malaysia (Kuala Lumpur), and India (Mumbai) regions on both the China site (aliyun.com) and the International site (alibabacloud.com).

China East 2 Finance

Malaysia (Kuala Lumpur)

India (Mumbai)

Create an ACK managed cluster

Node removal

Nodes can be removed from an ACK cluster. You can also choose whether to release the related ECS instances.

All regions

Remove a node

DaemonSet

DaemonSets are supported. DaemonSet is a daemon process that ensures that each node runs one replica of a pod.

All regions

No impact on workloads

Istio

Custom Istio Ingress and Egress gateways are supported by configuring different parameters.

All regions

Service Mesh

Istio CoreDNS

Istio CoreDNS is supported. You can use the CoreDNS plug-in to read Istio service entries and associate the IP addresses of the services to their host addresses.

All regions

Service Mesh

Existing ECS instances selectable during cluster creation

Existing ECS instances can be added as worker nodes when you create an ACK managed cluster.

All regions

Create an ACK managed cluster

November 2018

Feature

Description

Region

References

New region

ACK managed clusters are available in the Indonesia (Jakarta) region on the International site (alibabacloud.com).

Indonesia (Jakarta)

Create an ACK managed cluster

Terway

The Terway plug-in is released. Terway enables direct communication between containers through ENIs and provides higher network performance than Flannel.

All regions

Work with Terway

Thumbnail images for worker nodes

Thumbnail images are used to display the performance metrics of worker nodes, which makes it easy for you to view the status of nodes.

All regions

No impact on workloads

Node adding

Multiple existing nodes can be added to a cluster at the same time.

All regions

No impact on workloads

Rolling renewal of cluster certificates

Rolling renewal of cluster certificates is supported to prevent certificates from expiring.

All regions

No impact on workloads

October 2018

Feature

Description

Region

References

New region

ACK is available in the China South 1 Finance region on Alibaba Finance Cloud.

China South 1 Finance

Create an ACK dedicated cluster (discontinued)

New regions

No impact on workloads

Regions outside China

Create an ACK managed cluster

Deployment

Version management and rollback are supported for Deployments.

All regions

No impact on workloads

Istio

Istio is deeply integrated into ACK and Istio add-ons are supported.

All regions

No impact on workloads

September 2018

Feature

Description

Region

References

Kubernetes 1.11

  • Kubernetes 1.11 is supported to provide features, such as CRD update, CoreDNS general availability (GA), pod priority settings, and preemptive scheduling.

  • Multiple Kubernetes versions are supported, such as Kubernetes 1.10 and 1.11.

  • Multi-container applications and stateful applications are supported in the console.

All regions

Use a StatefulSet to create a stateful application

Container Registry

Images can be pulled from the private repositories of Container Registry without a password.

All regions

Auto scaling

Auto scaling of nodes is supported. ACK provides the auto scaling component for nodes to automatically scale in and out. Regular instances, GPU-accelerated instances, and preemptible instances can be automatically added to or removed from an ACK cluster as required. This feature is applicable to instances that are deployed across multiple zones and diverse instance types, and also supports different scaling modes.

All regions

Enable node auto scaling

Preemptible instances are supported.

No impact on workloads

All regions

August 2018

Feature

Description

Region

References

ACK managed clusters

ACK managed clusters are released for public preview.

All regions

Create an ACK managed cluster

Istio

Istio add-ons are supported.

All regions

No impact on workloads

July 2018

Feature

Description

Region

References

New regions

No impact on workloads

Australia (Sydney)

Create an ACK dedicated cluster (discontinued)

Canary releases and phased releases are supported.

No impact on workloads

All regions

No impact on workloads

June 2018

Feature

Description

Region

References

New regions

No impact on workloads

Japan (Tokyo)

China (Hohhot)

Create an ACK dedicated cluster (discontinued)

FPGA and HugePages are supported by Kubernetes 1.10.

No impact on workloads

All regions

No impact on workloads

Application monitoring and alerting

Application monitoring and alerting are supported.

All regions

No impact on workloads

Subscription supported when you create an ACK cluster

No impact on workloads

All regions

Create an ACK dedicated cluster (discontinued)

Ingresses and the exec and attach commands supported

No impact on workloads

All regions

Features

May 2018

Feature

Description

Region

References

New region

ACK is available in the China East 2 Finance region on Alibaba Finance Cloud. Alibaba Finance Cloud provides services in compliance with security regulations.

China East 2 Finance

Create an ACK dedicated cluster (discontinued)

ACK Serverless released

No impact on workloads

All regions

Create an ACK Serverless cluster

Blue-green releases, canary releases, and A/B testing supported

No impact on workloads

All regions

None

April 2018

Description

Region

References

ACK is available in five regions in Southeast Asia, the Middle East, and India. Kubernetes 1.9 is stably supported.

Malaysia (Kuala Lumpur)

Indonesia (Jakarta)

Singapore

India (Mumbai)

UAE (Dubai)

Create an ACK dedicated cluster (discontinued)

MySQL, RDS, RabbitMQ, and Spark are supported in Service Catalog.

All regions

This feature is phased out.

Management of applications released by using Helm is supported in App Catalog.

All regions

Manage releases by using Helm

March 2018

Feature

Description

Region

References

Kubernetes 1.9

Kubernetes 1.9.3 is supported. ACK releases Workloads API. By default, CRD is enabled. GPU scheduling is supported. You can select custom ECS images when you create a cluster. You can also reset images when you add nodes to a cluster.

All regions

No impact on workloads

Helm

App Catalog is released to allow you to deploy applications by using Helm.

All regions

Manage releases by using Helm

ServiceBroker

App Catalog is released to support ServiceBroker.

All regions

This feature is phased out.

CloudMonitor

Nodes can be monitored by using CloudMonitor.

All regions

Basic resource monitoring

January 2018

Feature

Description

Region

References

ACK and Container Registry released on the International site (alibabacloud.com)

No impact on workloads

Regions outside China

What is ACK?

Kubernetes 1.8.4 is supported to provide features such as security enhancement and auto scaling.

No impact on workloads

All regions

Enable node auto scaling

FlexVolume

The FlexVolume plug-in is released to support disks, NAS file systems, and OSS buckets.

All regions

Disk volumes, NAS volumes, and OSS volumes

Network policies and bandwidth throttling

Kubernetes network policies and bandwidth throttling are supported. This improves network performance.

All regions

Use annotations to configure load balancing

EBM instances supported

No impact on workloads

All regions

No impact on workloads

October 2017

Feature

Description

Region

References

Kubernetes 1.8.1

Kubernetes 1.8.1 is supported.

All regions

What is ACK?

Blockchain solutions released for public preview

No impact on workloads

All regions

No impact on workloads

August 2017

Feature

Description

Region

References

Kubernetes 1.7.2

No impact on workloads

All regions

Create an ACK dedicated cluster (discontinued)