Migration Process

Evaluation

Collect the following information about the cached data to be migrated (based on Information to be collected for the migration):

• Number of instances
• Number of databases (DBs) configured for each instance
• Number of keys in each DB
• DBs used for your services
• Space occupied by each instance
• Redis version
• Redis instance configurations (single-node, master/standby, or cluster)
• Mapping relationships between your services and instances

redis-cli -h ${redis_address} -p ${port}

• Run the following command to query the data distribution and obtain the IDs of DBs with data and the number of keys in each DB:

  info keyspace

  Query and record the number of keys in each DB for subsequent migration verification.

• Run the following command to query the space occupied by the instance data. Check whether the available disk space of Elastic Cloud Servers (ECSs) is sufficient for transition, and whether the instance specifications and remaining available memory are sufficient.

  info memory

  The occupied space can be obtained from the value of used_memory_human.

Preparation

After completing the evaluation, prepare the following items:

1. Mobile storage devices

   These devices are used to copy and transfer data in case of network disconnection (in scenarios with data centers of enterprises).

2. Network resources

   Create VPCs and subnets based on service planning.

3. Server resources

   Apply for ECSs to bear Redis clients. The ECSs are used to export or import cached data.

   Recommended ECS specifications are 8 vCPUs | 16 GB or higher.

4. DCS instances

   Create DCS instances based on the migration planning. If the number of instances exceeds the default quota, submit a service ticket or contact customer service.

5. Related tools

   Install the FTP tool and Redis migration tools.

6. Information to be collected

   Collect the contact information of people involved in the migration, server addresses, login credentials, cache instance information, and DB information.

7. Overall migration plan

   Formulate the overall migration plan, including the personnel arrangement, rehearsal, migration, verification, service switchover, and rollback solutions.

   Break down each solution into executable operations and set milestones to mark the end of tasks.

Rehearsal

The rehearsal phase aims to:

1. Verify the feasibility of the migration tools and migration process.
2. Discover problems that may occur during migration and make effective improvements.
3. Evaluate the time required for migration.
4. Optimize the migration steps and verify the feasibility of concurrent implementation of some tasks to improve migration efficiency.

Backup

Before migration, back up related data, including but not limited to cached data and Redis configuration files, in case of emergency.

Migration

After conducting one or two rounds of migration rehearsal and solving problems found in the rehearsal, start data migration.

Break down the migration process into executable steps with specific start and end confirmation actions.

Data Verification

Check the following items:

• The key distribution of each DB is consistent with the original or expected distribution.
• Main keys.
• Expiration time of keys.
• Whether instances can be normally backed up and restored.

Service Switchover

1. After the data migration and verification, use the new instances for your services.
2. If DB IDs are changed, modify the ID configurations for your services.
3. If your services are migrated from data centers or cloud platforms provided by other vendors to Huawei Cloud as a whole, services and cached data can be migrated concurrently.

Service Verification

After the service switchover:

1. Verify the connectivity between your service applications and DCS instances.
2. Verify whether cached data can be normally added, deleted, modified, and queried.
3. If possible, perform pressure tests to ensure that the performance satisfies the peak service pressure.

Rollback

If your services are unavailable after the data migration because unexpected problems occur and cannot be solved in the short term, roll back your services.

Since source Redis data still exists, you only need to roll back your services and use the source Redis instances again.

After the rollback, you can continue to restart from the rehearsal or even preparation phase to solve the problems.

Information to be collected for the migration

The following table lists the information to be collected in the evaluation and preparation phases.