Hologres:Multi-zone deployment across three availability zones

Introduction

Hologres's Multi-zone deployment feature extends the availability of an instance from a single availability zone to three availability zones within the same region. It provides fault isolation at the availability zone level to prevent a single-zone failure from interrupting your business services. You can use this feature to handle scenarios such as carrier network failures or compute infrastructure failures in a single availability zone, which enhances your business's disaster recovery capabilities.

Hologres Multi-zone deployment includes storage disaster recovery across three availability zones and high availability for compute across multiple availability zones:

• Storage disaster recovery across three availability zones: Also known as Zone-redundant Storage, this feature replicates an instance's data across multiple availability zones within a single region. The underlying availability zones are managed by the Pangu distributed storage system, and you do not need to manage their specific locations. If the storage in one availability zone becomes unavailable, the system automatically accesses a replica in another availability zone. This process requires no manual intervention and ensures continuous data access, providing zone-level disaster recovery for storage.

• High availability for compute nodes across multiple availability zones: If an instance uses Zone-redundant Storage, you can manually fail over its compute nodes to a healthy availability zone if a compute infrastructure failure occurs. This capability enhances the high availability of the compute layer, provided that the target availability zone has sufficient compute resources.

How it works

• Locally Redundant Storage: When an instance uses Locally Redundant Storage, it stores data in a single availability zone within a region. If this availability zone becomes unavailable, the data becomes inaccessible, and high availability for storage and compute cannot be achieved.

• Zone-redundant Storage: When an instance uses Zone-redundant Storage (for Multi-zone deployment), its data is replicated across multiple availability zones within the same region. If one availability zone becomes unavailable, Zone-redundant Storage ensures continuous data access, providing zone-level disaster recovery for storage and enabling it for compute.

Compared to Locally Redundant Storage, Zone-redundant Storage offers higher availability and more resilient disaster recovery capabilities, but it also increases storage costs. For details, see Billing overview. Other costs remain unchanged.

Multi-zone deployment provides storage disaster recovery and compute disaster recovery. The following sections describe how they work.

Purchase an instance with Multi-zone deployment

When you purchase a new instance, set the Storage Redundancy Type to ZRS.

After creating the instance, go to the Instance Details page. In the Storage Resources section, verify that the Storage Redundancy Type is set to Zone-redundant Storage (ZRS). If an availability zone failure occurs, follow the instructions in Disaster recovery guide. You can use this type of instance in the same way as an instance that uses Locally Redundant Storage.

Storage disaster recovery

If a storage failure occurs in the availability zone hosting your instance, Hologres notifies you by SMS or email. The system then begins automatic recovery:

1. Hologres automatically fails over the storage to another availability zone. No action is required from you, and your services recover automatically.

2. After the failover, data writes continue in a healthy availability zone, unaffected by the failed zone. The system automatically routes queries to the nearest available data replica. No application changes are required. If any jobs failed during the incident, you must rerun them.

3. Monitor your business operations to ensure that they are fully restored.

Compute disaster recovery

If your instance uses Zone-redundant Storage, Hologres lets you manually switch the compute availability zone to achieve high availability for compute nodes and quickly restore your services.

If a failure occurs in the compute availability zone of your instance, Hologres notifies you by SMS or email. Then, follow these steps to manually recover your compute resources:

1. Go to the Hologres console. On the Instances page, click the target instance's ID to open the Instance Details page.

2. In the left-side navigation pane, click Backup and Disaster Recovery and then click the Zone-disaster Recovery tab.

3. In the Compute Disaster Recovery section, click Switch Computing Zone.

4. If the target availability zone has sufficient compute resources, select a Computing Zone for Disaster Recovery in the Switch Computing Zone dialog box and click OK to migrate the compute nodes.

   After the migration is complete, the instance's basic configuration, such as its endpoint, remains unchanged. When the instance status changes to Running, rerun any failed jobs and monitor your business operations until they are fully restored.