Data Shard Management
Distributed Redis instances can be managed through the console details - shard management to manage the shards of the cluster. You can view the usage of each shard in the cluster instance and perform single shard management monitoring, the entry to shard management:
Shard Upgrading and Downgrading
Distributed Redis instances support configuration upgrades and downgrades for individual shards; if you need to upgrade or downgrade the capacity of a shard, you can select the operation item of the shard to upgrade or downgrade the shard:
Note: In actual production environments, once a single shard's memory usage is full, some data may not be written. Therefore, even if the entire cluster's memory capacity is sufficient, when a certain shard has a high load or high memory usage rate, it is necessary to scale up that shard. Additionally, it is recommended that users try to maintain load balance among all shards, avoiding situations where some shards are very idle, resulting in resource wastage, while others are heavily loaded.
When upgrading and downgrading a distributed Redis shard through the console, if the shard type is a physical machine type, there is no need to migrate data when expanding if the host resources are sufficient; however, if the host resources are insufficient, data migration is required. The impact on Redis is as follows: the Redis service remains available during upgrading and downgrading, but the load will increase when data synchronization begins, and there will be about 20 seconds of interruption during master-replica switching. Please try to execute these operations during low business periods. For Redis shrink operations, there is no impact on Redis throughout the process. If the shard type is a quick-featured host, the impact on Redis during expanding and shrinking is as follows: the load will increase when data synchronization begins, and there will be about 20 seconds of interruption during master-replica switching. Please try to execute these operations during low business periods.
Shard Monitoring and Management
Through the monitoring information of each individual shard, you can grasp the load situation of each shard and perform targeted management such as upgrading and expansion according to needs:
Moreover, each individual shard is by default connected to the "Default Cloud Memory Storage Alarm Template", and an alarm mechanism is triggered when the memory usage reaches the threshold set in the alarm template.
Shard Splitting Operation
Distributed version Redis users can perform shard splitting operations on the Shard Management page, with the option to split immediately and during maintenance times:
