ApsaraDB for Redis provides cluster version instances, allowing you to easily break through Redis’ single thread bottleneck. This is the preferred solution for businesses that require large Redis capacity or high performance. The ApsaraDB for Redis cluster version has built-in data partitions and reading algorithms. The overall process is transparent to users, saving you from the headache of developing and operating Redis clusters.
The ApsaraDB for Redis cluster version is composed of three components: proxy servers (service proxy), partition servers, and configuration servers.
In single-node configuration, the cluster consists of multiple proxies. The system automatically performs load balancing and failover for the cluster.
Each partition server is built on dual-copy highly available architecture. When the master node fails, the system automatically switches to the standby node to guarantee the high availability of the service.
Configuration servers are used to store cluster configuration information and partition policies. These servers currently adopt dual-copy high availability architecture, to guarantee high service availability.
- The quantities and configurations of these three components are specified by the system when you purchase the cluster version of the corresponding specification. Now, you cannot freely modify these options. The specifications are shown as follows:
|Cluster version specification||Proxy count||Partition server count||Memory size of each partition server|
|16 GB cluster version||8||8||2 GB|
|32 GB cluster version||8||8||4 GB|
|64 GB cluster version||8||8||8 GB|
|128 GB cluster version||16||16||8 GB|
|256 GB cluster version||16||16||16 GB|
Redis cluster versions expose a unified access domain name. You can visit this domain name for normal Redis access and data operations. Proxy servers, partition servers, and configuration servers do not support domain name access, so you cannot directly connect to them to perform operations.
You can purchase new cluster version instances or upgrade standard version (master/slave replication) instances to cluster version.
Large data volumes
Redis cluster version instances can be effectively scaled to accommodate different data volumes. Compared to the standard version, the cluster version can support larger capacities of 64, 128, and 256 GB to effectively meet your data scaling needs.
High QPS load
Standard version Redis instances cannot support high QPS and require you to use multi-node deployment to combat Redis’ single thread performance bottleneck. Redis cluster version instances support 16, 32, 64, 128, and 256 GB configurations and 8-node and 16-node deployment modes. Cluster version instances can increase the QPS load by 8 or 16 times compared with standard version instances.
Compared to standard version instances, Redis cluster version instances have looser restrictions on intranet throughput. This provides great support for businesses that read hotspot data and have high throughput requirements.
Apps insensitive to Redis protocol
Because of the multiple components, the cluster version architecture provides more limited support for Redis protocol than the standard version. For more information, see Supported Redis commands.