Tair (Redis OSS-compatible) instance architectures - Tair (Redis® OSS-Compatible)

Tair (Redis OSS-compatible) supports the standard, cluster, and read/write splitting architectures. You can select an architecture based on your business requirements.

Architectures

The following table describes the supported instance architectures. You can click the link to an instance architecture to view details about the architecture.

Instance architecture	Architecture diagram	Scenario
Standard master-replica architecture The master-replica architecture is used to ensure high availability.		Support for more native Redis features. Persistent storage on Redis instances. Stable query rate on a single Redis instance. Use of simple Redis commands, where only a few sorting and computing commands are required.
Cluster architecture In the cluster architecture, each data shard contains one master node and multiple replica nodes.		Large data volume. High queries per second (QPS). Throughput-intensive workloads.
Cluster architecture (read/write splitting) In the cluster architecture, read/write splitting is enabled for each data shard.		You can enable the read/write splitting feature for a cluster instance when the number of read requests to the cluster instance exceeds the performance limit of the master node.
Standard architecture (read/write splitting) The standard read/write splitting architecture consists of the master and replica nodes and multiple proxy nodes.		High QPS. Support for more native Redis features. Persistent storage on Redis instances.

Documentation applicability

To understand Tair (Redis OSS-compatible), make sure that you are familiar with the following concepts: architectures, editions such as Redis Open-Source Edition and Tair (Enterprise Edition), series types such as DRAM-based instances, and engine versions such as Redis 4.0 and 5.0. The descriptions and topics described in the preceding table are applicable to corresponding editions, series types, and engine versions. For information about the standard architecture, see Standard master-replica instances. This rule also applies to cluster and read/write splitting instances.