Tair is an in-memory database service based on ApsaraDB for Redis Community Edition and brings enterprise-level features and multiple storage media with cost-effectiveness.
DRAM-based instance: delivers ultra-high throughput and triples the performance of open source Redis, which is suitable for industries such as livestreaming, e-commerce, and online education.
Persistent memory-optimized instance: implements real-time data persistence at 70% the cost of ApsaraDB for Redis Community Edition, which is an ideal choice for the gaming industry.
ESSD-based instance: provides storage capacities of up to hundreds of terabytes, which is suitable for the storage of large amounts of warm and cold data with high performance.
Full Compatibility with Redis and Instance Upgrade Without Downtime
Tair is fully compatible with Redis. Migration to and from both databases can be performed out-of-the-box without any code modifications. Additionally, Tair is bundled with a variety of extended data structures that can simplify data storage and development complexity. The configuration change feature of ApsaraDB for Redis Community Edition can be used to upgrade an ApsaraDB for Redis instance to a Tair instance without downtime. View Details
Tair persistent memory-optimized instances use persistent memory and support command-level real-time persistence, semi-synchronization, and data restoration to a specific point in time. This instance series provides 90% read and write capabilities of ApsaraDB for Redis Community Edition at only 70% of the cost. This makes the instance series an ideal choice to replace solutions that use open source Redis for caching and ApsaraDB RDS for persistent storage. This way, the simplified database system ensures data consistency between caching and persistent data storage. View Details
Data Restoration to a Specific Point in Time
Tair provides the data flashback feature that can restore data to a specific point in time, accurate to the second. This feature enables you to restore an entire instance or specific keys to a new instance or the current instance of a specific point in time. This fine-grained data restoration capability prevents data loss caused by accidental operations and ensures data security. View Details
Global Distributed Cache and Data Synchronization
Global Distributed Cache is an in-house active geo-redundancy database system. The cloud-native geo-replication capability of Tair supports data synchronization across up to three regions. This helps you implement geo-redundancy, disaster recovery, data consistency across regions, and nearby access with reduced latency. View Details
Multiple Extended Data Structures
Tair provides a wide range of built-in extended data structures on top of the support for all open source Redis data structures. These extended data structures accelerate development timelines and improve code security by sparing you large amounts of code and the use of Lua scripts. Additionally, these data structures enable you to implement high-performance distributed locks, multidimensional leaderboards, sliding windows, and JSON file storage with ease. View Details
Ease of Use: A Broad Set of Data Structures to Set up Real-time Scenarios
Provides multiple instance series tailored towards hot, warm, and cold data storage, high performance, low costs, and data persistence.
Tair DRAM-based Instances
Uses a multi-threading model to increase the queries per second (QPS) of each node and handle a large number of connections with three times performance of open source Redis for requirements on high concurrency and low latency. View Details
Tair Persistent Memory-optimized Instances
Provides native command-level real-time data persistence and high performance for storage of large amounts of data with cost-effectiveness (90% throughput and 70% costs compared with ApsaraDB for Redis Community Edition). View Details
Tair ESSD-based Instances
Leverages Alibaba Cloud ESSDs to cater for massive data storage and query with up to 85% cost reduction compared with ApsaraDB for Redis Community Edition, and provides native Redis command-level data persistence and compatibility with native Redis data structures and APIs. View Details
Multiple Architectures: Standalone, Master-replica, Read/Write Splitting, and Cluster
Standalone for caching-only scenarios, master-replica for automatic failover, read/write splitting for read-intensive scenarios, and cluster for capacity-sensitive scenarios
Designed for caching-only scenarios and scales out smoothly to deliver cost-effective, high QPS performance.
Provides a master node and a replica node for real-time data synchronization. Automatic failover completes within seconds, and is completely transparent, ensuring the availability of your applications.
Adopts a distributed architecture in which each shard contains a master node and a replica node to implement automatic failover and ensure high availability, provides multiple specifications, and supports push-button scaling.
Read/Write Splitting Architecture
Designed for read-dominant scenarios with a maximum of 100,000 writes per second and 600,000 reads per second to break bottlenecks of hotkey reads.
Enterprise-level Capabilities: Global Distributed Cache, Data Flashback, and Multiple Data Structures
Enterprise-level features for fast implementation in various scenarios
Data Restoration to a Specific Point in Time
Allows you to restore data to a specific point in time within the last seven days by second to prevent data loss caused by accidental operations and improve data security. View Details
Real-time Data Persistence
Implements data persistence without dependence on append-only files (AOF) and Redis Database (RDB) files to prevent data loss from power outages and enhance data reliability. View Details
Global Distributed Cache
Supports real-time read and write synchronization across zones and regions, provides cross-region replication, and achieves geo-redundancy, disaster recovery, and nearby access with reduced latency. View Details
Handles traffic spikes and prevents out-of-memory errors to ensure business stability.
Accelerated Hotkey Read
Uses proxy nodes to cache hotkeys and resolve hotkey-related performance bottlenecks. You no longer have to manually modify your code or predict resource usage. View Details
Enhanced Data Structures
Provides Redis-compatible enhanced data structures to implement multidimensional leaderboards, distributed locks, Bloom filters, and trajectory monitoring for agile development. View Details
Elasticity: Multidimensional Scalability and Transparent Upgrades
One-click configuration change to improve instance capacities and performance or change instance architectures
Provides a wide range of memory configurations, allowing you to scale as your business changes.
Supports memory and throughput scaling to achieve tens of millions of queries per second.
Flexibly switches between standalone and master-replica architectures for in-memory storage or persistent storage capabilities.
Intelligent O&M: Purpose-built Monitoring and Management Platforms Provide Push-button Troubleshooting
Provides user-friendly graphical database management and O&M platforms. This helps reduce O&M costs and support real-time and historical analysis of large keys and hotkeys.
Monitors instance metrics such as CPU utilization, memory usage, number of connections, and disk usage in real time and raises alerts to track your instance status.
Visualized Management Platform
Clones, backs up, and restores instances with one click.
Visualized DMS Platform
Provides the Data Management (DMS) console for visualized management and enhanced development and maintenance efficiency.
Powerful Monitoring Capabilities
Provides monitoring within seconds for storage and latency features such as real-time hotkeys and large keys, latency insights, and audit logs.
Comprehensive Diagnostic Reports
Provides checks for request latency and instance health with automatic or manually generated diagnostic reports to identify issues.
Livestreaming applications constantly experience traffic spikes. Tair cluster instances can be used to meet the needs for high concurrency, low latency, and up to millions of QPS.
Transparent Elastic Scaling
Handles system access, user visits, and traffic spikes without affecting your business.
Data Security and Stability
Each shard in a cluster instance uses a master-replica architecture to implement disaster recovery and ensure data security.
Meets the needs for duplicate servers, cross-server data transfer, and elastic scaling.
Gaming applications have high requirements for simplified storage solutions, scalability, and high performance. Tair effectively covers these needs so you can focus on developing your applications.
Gaming applications have requirements for duplicate servers, cross-server data transfer, and elastic scaling. In this case, data persistence can decrease the load on the database engine and simplify development.
Meets the needs for duplicate servers of gaming applications and allows you to clone instances by using backup files.
Provides free seven-day retention period and one-click restoration for backup files to ensure data security.
Seamless Integration with Big Data Computing
E-commerce promotions such as flash sales exert heavy load on the inventory management system. In this case, Tair can handle read requests with ease and provides data persistence.
Uses a multi-threading model to handle up to 300,000 QPS and high concurrency scenarios.
Provides point-in-time snapshots for AOFs and RDB files to ensure data persistence.
Optimized Storage Architecture
Uses a storage architecture that combines Tair with ApsaraDB RDS and provides shared caching and data persistence storage to improve database performance.
Web applications are read-and-write-intensive and cost-sensitive. Tair standalone instances meet these requirements with caching capabilities, high QPS, and comprehensive data persistence.
High Read and Write Performance
Tair cluster instances that use the multi-threading model can handle up to 300,000 QPS and massive traffic.
Protects your business from single-node failure.
Tair standalone instances deliver cost-effective, caching-only performance.