DCS Redis single-node cluster instances adopt the proxy cluster mode, and the data shards adopt a single-node architecture. These instances break the single-thread bottlenecks of Redis and meet the requirements of large-capacity or high-performance business scenarios and are more cost-efficient compared with the active/standby cluster instances.
Note:
l Since the Redis single-node cluster data shards adopt a single-node architecture and do not have HA and service continuity, it is not recommended that you use it in a production environment. But if you do, please be sure to note the risks.
l Single-node cluster instances can be upgraded to active/standby cluster instances. For details, see Type Change.
l When a node of a data shard fails, the system initiates a new Redis process to ensure high service availability, and user data is restored based on the default automatic persistence files to achieve business continuity.
Architecture Diagram
Architecture Description
l Load balancer: It works in the active/standby HA mode to receive client requests. The Redis cluster instance provides the IP address and domain name for access, namely, the load balancer address.
l Proxy: A Redis cluster proxy server evenly distributes client requests to available nodes based on the load of each Redis node to achieve load balancing. At the same time, it monitors the health status of Redis nodes and automatically forwards requests to other available nodes to achieve failover and reduce the impact of single-node faults.
l Redis data shards: Each data shard uses a single-node architecture without HA. When a node of a data shard fails, the system initiates a new Redis process to ensure service HA. If data is not backed up, the data on faulty nodes will be lost.
Application Scenarios
Cache-Only Scenarios
Frequently accessed data can be stored in the Redis cluster memory to speed up data reading. Each data shard uses a single-node deployment architecture. When the node fails and an automatic switchover is performed, data will be lost. It is recommended to use active/standby cluster instances if you have high requirements for data reliability.
Throughput-Intensive Application Scenarios
For throughput-intensive scenarios, Redis cluster can meet the high performance and high concurrency requirements with its features such as distributed architecture, horizontal scaling, and concurrent processing.
Scenarios with High QPS Pressure
A Redis cluster is deployed on multiple nodes to break down the performance bottlenecks in single threads, and can better support scenarios with a high QPS.
Scenarios Insensitive to the Redis Protocol
The cluster version architecture has introduced multiple components and has certain limitations in terms of support for the Redis protocol compared to the standard version instances. Therefore, it is not suitable for scenarios that require high Redis protocol compatibility.