HA Solution on the Cloud

The availability of workloads on the public cloud is determined by the availability of the application layer, architecture design, and cloud services. The availability of workloads on the public cloud is a systematic project. The reliability of workloads on the public cloud depends on the availability design of the overall service architecture and O&M specifications (such as the backup mechanism, routine drills, and personnel operation specifications).

Figure 1 Service availability solution

Most Huawei Cloud services can be deployed in HA mode. HA capabilities are implemented for data centers, hardware, data, and self-service functions. Huawei Cloud data centers are deployed around the globe to meet resource requirements for different regions. Each region is divided into multiple AZs. Each AZ has independent cooling, fire extinguishing, moisture-control, and electrical facilities, and the failure of one AZ does not affect others. There are four types of HA deployments:

• Single-AZ deployment: Generally, single-AZ deployment is not recommended on the cloud, except for latency-sensitive services. You can deploy cloud services in primary/standby or cluster mode to quickly recover services from a fault. Automatic detecting and switching over faulty nodes in a cluster can eliminate single points of failure (SPOFs).
• Dual-AZ (intra-city) HA: For services that require high availability, you can deploy services in multiple equipment rooms in the same city. This way, service continuity is guaranteed even if the network, physical device, or power supply of one equipment room fails. Huawei Cloud users can deploy services across AZs. AZs are isolated from each other, so if one AZ fails, you can switch services to another AZ to quickly recover services. Most cloud service products have capabilities related to dual-AZ HA. You only need to select desired capabilities when purchasing the cloud service.
• Geo-redundant DR: For some ultra-large or commercial systems that demand extra robust protection, the dual-AZ (intra-city) HA solution cannot solve regional-level faults, such as earthquakes and floods. So, you need to add a remote DR equipment room on the basis of intra-city DR. That way, you are protected in the event of a regional disaster.
• Cross-cloud HA: To meet requirements for multi-cloud HA, enterprises can deploy a primary site on Huawei Cloud and deploy a standby site on other cloud vendor's platform.