Ubiquitous Cloud Native Service (UCS)

UCS is the industry's first distributed cloud native product. It enables you to run cloud native applications across regions and clouds with intelligent traffic distribution, while enjoying a consistent experience in terms of the deployment, management, and ecosystem.

Unified Compute Scheduling

You can use metric-based scaling policies to deploy compute on Huawei Cloud and third-party clouds, as well as at edge sites and customer premises.

Simple Configuration Management

You can enable GitOps for declarative version management and automated delivery of Kubernetes native resources.

Intelligent O&M

There are 24/7 health monitoring, global monitoring, intelligent inspection, and standard O&M APIs for interconnection with third-party O&M systems.

Kubernetes Clusters That Can Be Connected and Managed on Any Infrastructure in Any Region

Cluster Type

Huawei Cloud Clusters

On-Premises Clusters

Multi-Cloud Clusters

Attached Clusters

Overview

Huawei Cloud CCE standard clusters and CCE Turbo clusters

Kubernetes clusters that are provisioned by UCS but running on your on-premises data center and are available for offline use

Kubernetes clusters that are provisioned by UCS but running on third-party clouds

Third-party Kubernetes clusters that comply with Cloud Native Computing Foundation (CNCF) standards

Deployment Mode

Cluster resources are fully hosted on Huawei Cloud to simplify service cloud migration.

Software is provided by Huawei Cloud but deployed in your on-premises data center so cloud and premises resources can work together.

Software is provided by Huawei Cloud but automatically deployed on third-party clouds.

Kubernetes clusters are deployed on third-party clouds, or self-managed Kubernetes clusters are deployed.

Application Scenarios

You are running workloads in Huawei Cloud clusters, you plan to migrate workloads to the cloud, or the workloads in your self-managed Kubernetes clusters need to be scheduled to the cloud during traffic surges.

You have an on-premises data center, your services require more robust data security and reliability, or you need flexible, elastic AI infrastructure to accelerate AI training and inference.

You have infrastructure resources from third-party cloud vendors and want to use Huawei Cloud UCS for multi-cloud management, so that you can deploy applications across clouds.

You have workloads running on third-party clouds or in self-managed Kubernetes clusters and want to use a multi-cloud platform for hybrid service deployment and higher application reliability.

Register Now

Huawei Cloud Clusters

Huawei Cloud CCE standard clusters and CCE Turbo clusters

Cluster resources are fully hosted on Huawei Cloud to simplify service cloud migration.

On-Premises Clusters

Kubernetes clusters that are provisioned by UCS but running on your on-premises data center and are available for offline use

Software is provided by Huawei Cloud but deployed in your on-premises data center so cloud and premises resources can work together.

Multi-Cloud Clusters

Kubernetes clusters that are provisioned by UCS but running on third-party clouds

Software is provided by Huawei Cloud but automatically deployed on third-party clouds.

You have infrastructure resources from third-party cloud vendors and want to use Huawei Cloud UCS for multi-cloud management, so that you can deploy applications across clouds.

Attached Clusters

Third-party Kubernetes clusters that comply with Cloud Native Computing Foundation (CNCF) standards

Kubernetes clusters are deployed on third-party clouds, or self-managed Kubernetes clusters are deployed.

Why Huawei Cloud UCS?

Efficient Cluster Management

Access from the public network, VPN, and Direct Connect is supported.
A dashboard is provided for the comprehensive view of global cluster resource and application distribution.
Consistent permission management is available for all clusters. Enterprise-level permission management is also supported.

Collaboration of Huawei Cloud and On-Premises Data Center

On-premises workloads can be scheduled to the cloud, and cloud resources can be scaled out within seconds to handle traffic bursts with ease.
Relying on Volcano and Huawei Cloud EulerOS 2.0, the hybrid deployment of online and offline services in on-premises clusters improves the resource utilization by 40%.
GPU virtualization enables GPU isolation and improves resource utilization during AI training and inference.

Efficient O&M

You get a comprehensive view of the clusters in every region, so they can be centrally monitored.
Intelligent inspection can automatically protect the health of clusters in all regions.
There is a standard O&M API that allows your O&M system to quickly interconnect with third-party O&M systems through REST APIs or Exporter.

Karmada-based Multi-Cluster Management

There are multiple scheduling policies such as ClusterName, labels, taints, and tolerations.
Cross-cluster failovers are supported. If a cluster becomes faulty, instances in the faulty cluster are automatically migrated to a healthy cluster.
Cross-cluster auto scaling is supported. Application instances can be provisioned based on cluster weights.
Traffic can be distributed across clusters by cluster weight, cluster priority, and automatic load balancing.

Countless Applications, Fast Service Deployment

Global Management
AI Computing
Modernized Application Delivery
Multi-Active DR

Global Management

Introduction

Business groups can use UCS to set up a cloud native digital platform to provide unified management, O&M, configurations, and ecosystem for clusters in all regions.

Pain Points

Inefficient Management

Business groups typically have complex infrastructures, with a number of branches spread out over various locations, which make managing them less efficient.

Slow Fault Response

The HQ and each branch use different insight and analysis tools. This impairs fault detection and makes them slower to respond to faults.

No Lightweight Cloud Migration Solution

The existing hardware and virtualization environments cannot be fully reused, so legacy investments go wasted.

Benefits

1. The management efficiency is greatly improved.

2. Full-stack cloud native enables more agile service development and testing and more convenient delivery and O&M.

3. Only three physical machines or VMs are needed to set up a cloud native container platform.

AI Computing

Introduction

A flexible, elastic AI container infrastructure is created based on UCS. This infrastructure leverages the performance optimization, Volcano scheduler, and GPU virtualization of Huawei Cloud EulerOS to provide an efficient, scalable computing platform for AI model training and inference. With this infrastructure, organizations can easily deploy AI models, accelerating the innovation of intelligent applications.

Pain Points

Low GPU Resource Utilization

In AI training and inference, GPU resources cannot be fully used or managed in a refined manner, so they often go wasted.

Training Scheduling and Deadlocks

Organizations often rely on the distributed training of compute-intensive models. However, distributed training can run into problems like excessive wait times, deadlocks, and intensive I/O loads. In this case, task scheduling needs to be optimized to improve efficiency.

Benefits

1. Huawei-developed GPU virtualization isolates compute from GPU memory, which improves resource utilization by 20%.

2. The Volcano scheduler provides scheduling algorithms like Gang scheduling and I/O aware scheduling to solve problems in distributed training.

Modernized Application Delivery

Introduction

Huawei Cloud CodeArts provides visualized, customizable continuous delivery pipelines. Based on the release rollback and version iteration control of GitOps, Huawei Cloud CodeArts shortens the delivery period from one week to one day, significantly accelerating application delivery and O&M in multi-cloud environments.

Pain Points

Complex Application Delivery

In multi-cluster scenarios, different configurations are required to deploy the same application on a public cloud, on premises, or at the edge.

Complex Application O&M

In multi-cluster scenarios, there is no unified approach to detect and control risks in advance.

Benefits

1. UCS works with Huawei Cloud CodeArts to provide out-of-the-box solutions.

2. GitOps is used for auditable and sustainable deployment.

3. Service observability allows you to identify risks in advance.

Multi-Active DR

Introduction

Multi-active DR provides cross-cloud load balancing, real-time data synchronization, and automatic failovers to ensure the system will not break down due to individual faults. This significantly improves system stability and fault tolerance.

Pain Points

Individual Faults

If an application running in a single data center or on a single cloud platform fails, services are interrupted, and user experience is affected.

Long Recovery Time Objective (RTO)

If a disaster strikes a single environment, it takes long time to recover the environment.

Unbalanced Resource Use

Some nodes excessively consume resources while others remain idle due to traffic fluctuation and improper configuration.

Benefits

1. High availability maintains service continuity even if there are individual faults.

2. Cross-cloud load balancing and data synchronization improve resource allocation and performance optimization.

3. Multiple cloud vendors or data centers reduce the reliance on a single vendor.

Documentation

Start Your UCS Journey

Buy

More Services

Cloud Container Instance (CCI)

Build applications efficiently with serverless containers

Cloud Container Engine (CCE)

Experience enterprise-class, managed Kubernetes service

SoftWare Repository for Container (SWR)

Manage your container images easily

Ubiquitous Cloud Native Service (UCS)