Huawei Cloud Stack 8.3.0 O&M Guide 04
RDS DR Management
Cross-Region DR Architecture
RDS Cross-Region DR Architecture
The following figure shows the RDS cross-region DR architecture.
- An RDS primary/standby instance can be deployed across AZs in a region. Data is synchronized between the primary and standby instances based on the primary/standby replication.
- An RDS primary/standby instance deployed across AZs uses a unified address for external access. If the primary instance is faulty, RDS fails over to the standby instance in seconds, with the access address unchanged.
- RDS also supports cross-region DR deployment. You can deploy an RDS instance in the production center and DR center, respectively. Both of them are kept running. If the instance in the production center fails, the instance in the DR center can take over services quickly.
Cross-AZ DR Architecture
RDS Cross-AZ DR Architecture
The following figure shows the RDS cross-AZ DR architecture.
- Cross-AZ HA supports switchover in seconds: RDS for MySQL supports cross-AZ deployment. The primary instance is deployed at the production site, and the standby instance is deployed at the intra-city DR site. One primary/standby switchover takes only several seconds.
- Data replication: Data is replicated from the primary instance to the standby instance. Both semi-synchronous and asynchronous replication are supported. To ensure data security, semi-synchronous replication is recommended. To improve performance, asynchronous replication is recommended. The default replication model between the primary instance and its read replicas is asynchronous and cannot be changed.
- Multiple read replicas: One RDS for MySQL instance supports up to five read replicas to share read traffic and improve the query throughput.
- Primary/standby switchover transparent to applications: A pair of primary/standby RDS for MySQL instances, deployed across AZs, uses the same address for external access. When a site becomes faulty, the address remains unchanged.
Planning and Design
This section describes the network planning and design for RDS cross-region DR. For more information, see Huawei Cloud Stack 8.3.0 Integration Design Suite.
Network Plane |
Description |
---|---|
DMZ_Service |
Used for communication among RDS management nodes. This network plane must communicate with the External_OM plane. |
DMZ_Service_Advance |
Used by the RDS management plane to manage RDS instances. The network planes DMZ_Service_Advance and External_Relay_Network must be accessible to each other so that the RDS management plane can manage RDS instances. |
External_OM |
DBS Platform management plane database and ZooKeeper nodes are deployed on this plane. This plane must communicate with the DMZ_Service plane. |
POD_Service_Cluster.cType |
VM management subnets of the database resource tenant. A CIDR block that does not overlap with the VPC subnet of the tenant is automatically selected from the two subnets for instance provisioning. |
POD_Service_Cluster.bType |
Local |
Local Region |
Local Plane |
Peer |
Peer Region |
Peer Plane |
Description |
---|---|---|---|---|---|---|
RDS |
Region |
DMZ_Service |
RDS |
Region |
External_OM |
Intra-region communication: RDS components access database components across planes. |
RDS |
Region |
DMZ_Service_Advance |
RDS |
Region |
External_Relay_Network |
Intra-region communication: The RDS management plane accesses the internal public network addresses of RDS instances to manage RDS instances. |
RDS |
Region |
DMZ_Service |
RDS |
Region |
DMZ_Service |
Cross-region communication: In cross-region DR of the RDS service plane, the management plane of the primary region exchanges data with that of the DR region. |
ManageOne |
Global |
DMZ_Service |
RegionLB |
Region |
OM_ECMP_VIP |
Intra-region communication and communication between Global and Region: The ManageOne static server communicates with the database service backend through RegionLB-LVS-Float-IP of RegionLB. |
ConsoleFramework |
Region |
DMZ_Service |
RegionLB |
Region |
OM_ECMP_VIP |
Intra-region communication: The static server communicates with the database service backend through RegionLB-LVS-Float-IP of RegionLB. |
RegionLB |
Region |
External_OM |
DBS Platform |
Region |
DMZ_Service |
Intra-region communication: The console accesses services provided by DBS Platform through RegionLB. |
DBS Platform |
Region |
DMZ_Service |
DBS Platform |
Region |
External_OM |
Intra-region communication: the DBS Platform component accesses database components across planes. |
DBS Platform |
Region |
DMZ_Service |
ManageOne |
Global |
External_OM |
Communication between Global and Region: DBS Platform in the Region zone communicates with ManageOne in the Global zone to report alarms and resources and access the ManageOne authentication and authorization interface. |
DBS Platform |
Region |
DMZ_Service |
HAProxy |
Region |
External_OM |
Intra-region communication: DBS Platform accesses basic cloud services to apply for resources such as VMs. |
DBS Platform |
Region |
DMZ_Service_Advance |
DBS Platform |
Region |
External_Relay_Network |
Intra-region communication: The database service management plane communicates with the ECSs where DB instances are located through the internal public network plane for DB instance management. |
Basic Resource Deployment and RDS DR Deployment
RDS DR Deployment Dependency
The following table lists the services on which RDS cross-region DR depends. Global-level services (such as ManageOne) on the cloud base management plane must be deployed in cross-region DR mode.
Service |
Dependency Type |
Scenario |
Description |
---|---|---|---|
ManageOne |
Strong |
Management plane |
Used to provide O&M capabilities. |
VPC |
Strong |
Data plane |
Used to provide intra-cloud network automation capabilities. |
DBS Platform |
Strong |
Management plane |
Used to provide common capabilities of database services. |
ReverseLB/RegionLB/PODLB |
Strong |
Management plane |
Used for logging in to ManageOne Maintenance Portal. The Nginx is used as the backend of RegionLB in the management zone. |
ConsoleFramework |
Strong |
Management plane |
Used to register the service console. |
Auto change platform |
Strong |
Management plane |
Used for deployment and upgrade. |
DRS |
Strong |
Management plane |
Used to synchronize data from the primary instance to the DR instance across regions in RDS cross-region DR. |
EIP |
Strong |
Data plane |
Used to connect the networks between the service plane of the primary region and the OBS of the DR region in RDS cross-region DR. |
ELB |
Strong |
Data plane |
Used for load balancing in a DC. |
Deploying Basic Resources
Install the dependent services by following the instructions provided in "Deployment Guide (DR Scenarios)" of Huawei Cloud Stack 8.3.0 Software Installation Guide and Huawei Cloud Stack 8.3.0 Software Installation Guide for gPaaS & AI DaaS Services.
Deploying the RDS Service
Install RDS by referring to "Relational Database Service (RDS) Installation Guide" in Huawei Cloud Stack 8.3.0 Software Installation Guide for gPaaS & AI DaaS Services.
RDS Instance Setup
Planning and Creating VPCs
Planning VPCs
Because the application layer and database layer are deployed in the same VPC, you only need to apply for one VPC in each DC.
Creating VPCs
- Log in to ManageOne as a VDC administrator using a browser.
URL in non-B2B scenarios: https://Domain name of ManageOne Operation Portal, for example, https://console.demo.com
URL in B2B scenarios: https://Domain name of ManageOne Tenant Portal, for example, https://tenant.demo.com
URL of the unified portal: https://Domain name of the ManageOne unified portal, for example, https://console.demo.com/moserviceaccesswebsite/unifyportal#/home. On the homepage, choose Self-service Cloud Service Center to go to ManageOne Operation Portal.
You can log in using a password or a USB key.- Login using a password: Enter the username and password.
The password is that of the VDC administrator.
- Login using a USB key: Insert a USB key with preset user certificates, select a device and certificate, and enter a PIN.
- Login using a password: Enter the username and password.
- Click
in the upper left corner of the page and choose Network > Virtual Private Cloud. The network console is displayed.
- Click Apply for VPC. In the displayed Select Service dialog box, click Apply Now.
- Click Next. On the Apply for VPC page, configure the basic information, subnet, and IP address.Figure 13-26 Creating a VPC in the production center
- Click Apply Now to complete the VPC creation.
- Repeat 2 to 5 to create a VPC in the DR center.
Planning and Creating EIPs
Planning EIPs
To ensure that the databases can be accessed by external systems and DRS can connect to the source database in the production center, you need to apply for one EIP for the production center and DR center, respectively.
Creating EIPs
- Log in to ManageOne as a VDC administrator using a browser.
URL in non-B2B scenarios: https://Domain name of ManageOne Operation Portal, for example, https://console.demo.com
URL in B2B scenarios: https://Domain name of ManageOne Tenant Portal, for example, https://tenant.demo.com
URL of the unified portal: https://Domain name of the ManageOne unified portal, for example, https://console.demo.com/moserviceaccesswebsite/unifyportal#/home. On the homepage, choose Self-service Cloud Service Center to go to ManageOne Operation Portal.
You can log in using a password or a USB key.- Login using a password: Enter the username and password.
The password is that of the VDC administrator.
- Login using a USB key: Insert a USB key with preset user certificates, select a device and certificate, and enter a PIN.
- Login using a password: Enter the username and password.
- Click
in the upper left corner of the page and choose Network > Elastic IP. The network console is displayed.
- Click Apply for EIP. In the displayed Select Service dialog box, click Apply Now.
- Click Next. On the Apply for EIP page, configure the basic information and bandwidth.Figure 13-28 Applying for an EIP in the production center
- Click Apply Now to create an EIP for the production center.
Creating RDS Instances
- Log in to ManageOne as a VDC administrator using a browser.
URL in non-B2B scenarios: https://Domain name of ManageOne Operation Portal, for example, https://console.demo.com
URL in B2B scenarios: https://Domain name of ManageOne Tenant Portal, for example, https://tenant.demo.com
URL of the unified portal: https://Domain name of the ManageOne unified portal, for example, https://console.demo.com/moserviceaccesswebsite/unifyportal#/home. On the homepage, choose Self-service Cloud Service Center to go to ManageOne Operation Portal.
You can log in using a password or a USB key.- Login using a password: Enter the username and password.
The password is that of the VDC administrator.
- Login using a USB key: Insert a USB key with preset user certificates, select a device and certificate, and enter a PIN.
- Login using a password: Enter the username and password.
- Click
in the upper left corner of the page and choose Database > Relational Database Service. The RDS console is displayed.
- Select a DB engine version, AZ, DB instance type, and time zone.
- Select a planned VPC and specify the administrator password.
- Confirm the settings.
- If you need to modify your settings, click Previous.
- If you do not need to modify your settings, click Submit.
- Repeat 1 to 5 to apply for an RDS instance in the DR center.
The instance class, storage space, and type of the DR instance must be the same as those of the primary instance to prevent DR data inconsistency or task failures.
- Perform the following steps to bind an EIP to the newly created RDS instance in the production center:
- On the Instances page, click the instance name to go to the Basic Information page.
- On the Connectivity & Security page, click Bind next to the EIP field.
- In the displayed dialog box, all EIPs in the unbound status are listed. Select the required EIP and click OK. If no available EIPs are displayed, click View EIP to obtain an EIP.
- On the EIPs page, view the EIP that has been bound to the DB instance.
- On the Instances page, click the instance name to go to the Basic Information page.
Establishing a DR Relationship
Creating a DRS Instance
- Log in to ManageOne as a VDC administrator using a browser.
URL in non-B2B scenarios: https://Domain name of ManageOne Operation Portal, for example, https://console.demo.com
URL in B2B scenarios: https://Domain name of ManageOne Tenant Portal, for example, https://tenant.demo.com
URL of the unified portal: https://Domain name of the ManageOne unified portal, for example, https://console.demo.com/moserviceaccesswebsite/unifyportal#/home. On the homepage, choose Self-service Cloud Service Center to go to ManageOne Operation Portal.
You can log in using a password or a USB key.- Login using a password: Enter the username and password.
The password is that of the VDC administrator.
- Login using a USB key: Insert a USB key with preset user certificates, select a device and certificate, and enter a PIN.
- Login using a password: Enter the username and password.
- Click
in the upper left corner and choose Database > Data Replication Service. The DRS console is displayed.
- On the Disaster Recovery Management page, click Create Disaster Recovery Task in the upper right corner.
- Select Current cloud as standby for Disaster Recovery Relationship, MySQL for Service DB Engine, and the RDS for MySQL instance created in the DR center for DR DB Instance, and click Submit.
After the MySQL DR link is set up, DRS automatically sets the DR instance to the read-only state. For details, see "Data Replication Service (DRS) 2.23.07.200 User Guide (for Huawei Cloud Stack 8.3.0)" > "Getting Started with Real-Time Disaster Recovery" > "Step 1: Create a DR Task" in the Data Replication Service (DRS) 2.23.07.200 Usage Guide (for Huawei Cloud Stack 8.3.0).
- Return to the Disaster Recovery Management page and click the refresh icon to view the newly created DR task.
Configuring a DR Task
- On the Disaster Recovery Management page, select the DR task and click Edit in the Operation column.
- Select the RDS for MySQL instance created in the production center as the source database and click Test Connection. If the message Test successful is displayed, the configuration is correct.
- Configure the username and password of the destination database and click Test Connection. If the message Test successful is displayed, the configuration is correct.
- After the connection test is successful, click Next. Ensure that the check success rate reaches 100%. Then, click Next and then Submit.
- Check whether the task status is Disaster recovery in progress.
- Click the name of the DR instance to view detailed DR information and confirm the DR progress and DR data.
Performing a Primary/Standby Switchover
If the instance in the production center is faulty, you can manually switch the roles of the instance in the production center and that in the DR center. This operation can be performed on ManageOne Operation Portal.
- Log in to ManageOne as a VDC administrator using a browser.
URL in non-B2B scenarios: https://Domain name of ManageOne Operation Portal, for example, https://console.demo.com
URL in B2B scenarios: https://Domain name of ManageOne Tenant Portal, for example, https://tenant.demo.com
URL of the unified portal: https://Domain name of the ManageOne unified portal, for example, https://console.demo.com/moserviceaccesswebsite/unifyportal#/home. On the homepage, choose Self-service Cloud Service Center to go to ManageOne Operation Portal.
You can log in using a password or a USB key.- Login using a password: Enter the username and password.
The password is that of the VDC administrator.
- Login using a USB key: Insert a USB key with preset user certificates, select a device and certificate, and enter a PIN.
- Login using a password: Enter the username and password.
- Click
in the upper left corner and choose . The DRS console is displayed.
- Select the DR task and choose Batch Operations > Primary/Standby Switchover to change the status of the RDS DR instance from read-only to read/write.
- Change the instance IP address on your application and connect to the instance using the new address. Data can be properly read from and written to the instance.