Tair (Redis® OSS-Compatible):What is Tair (Redis OSS-compatible)?

Instance type

Description

Redis Open-Source Edition

A high-performance in-memory database product that is compatible with Redis. It supports architectures such as standard (primary-replica), cluster, and read/write splitting.

Tair (Enterprise Edition)

Tair (Enterprise Edition) is an enhanced Redis service developed based on Redis Open-Source Edition. Based on storage media such as dynamic random-access memory (DRAM), non-volatile memory (NVM), and enterprise SSDs (ESSDs), Tair offers various product forms that consider the three core dimensions of access latency, persistence requirements, and overall cost. Tair provides stronger performance, more data structures, and more flexible storage methods to meet business requirements in different scenarios.

• Memory-optimized: Uses a multi-threaded model, integrates some features of Alibaba's Tair, and supports multiple Tair data structures. It is highly suitable for specific business needs.

• Persistent memory-optimized: Based on persistent memory technology, it provides a large-capacity, Redis-compatible, in-memory database product. Data persistence does not rely on traditional disks. It ensures the persistence of each operation while providing throughput and latency close to Redis Open-Source Edition, greatly improving business data reliability.

• Disk-based: Developed based on ESSDs and SSDs, it is compatible with core Redis data structures and interfaces. The cost is as low as 15% of Redis Open-Source Edition, and the performance is about 60% of Redis Open-Source Edition. It can provide large-capacity, low-cost, and highly persistent database services. It is suitable for warm and cold data storage scenarios that require Redis compatibility, large capacity, and high access performance.

Architecture type

Description

Standard architecture

Data is synchronized from the primary node (Master) to the replica (Replica) in real time. If the primary node fails, the system quickly switches services to the replica node. The entire process is automatic and does not affect your business, ensuring high service availability.

Cluster architecture

Cluster instances use a distributed architecture. Each data shard supports primary/secondary failover (master-replica), which enables automatic disaster recovery and fault migration to ensure high service availability. Multiple instance types are available. You can select an instance type based on your business workload and freely upgrade or downgrade the instance type as your business grows. The cluster architecture supports two connection modes:

• Proxy mode (recommended): Provides intelligent connection management to reduce application developer costs.

• Direct connection mode: Allows clients to bypass proxy servers and directly access backend data shards. This reduces network overhead and service response time. This mode is suitable for businesses that require extremely high Redis response speeds.

Read/write splitting

Read/write splitting instances use a primary-replica (Master-Replica) architecture to achieve high availability. The primary node attaches read replicas (Read Replica) to replicate data and supports linear scaling of read performance. Read replicas can effectively mitigate performance issues caused by hot keys and are suitable for business scenarios with high read/write ratios. Two versions of read/write splitting instances are available.

• Read/write splitting (cloud-native edition): Read-only nodes synchronize data from the primary node in a star replication architecture. You can customize the number of read-only nodes (1 to 4 per shard for cluster architecture, 1 to 9 for standard architecture). This is suitable for large-scale business scenarios with high read/write ratios.

• Read/write splitting (classic edition, discontinued): Read-only nodes use a chained replication architecture. You can configure 1, 3, or 5 read-only nodes.