ApsaraDB for HBase is a distributed NoSQL database that stores data in key-value pairs and wide tables. It delivers millisecond query responses, handles tens of millions of queries per second (QPS), and scales to petabytes (PB) of data — making it well-suited to high-concurrency, large-scale workloads across multiple industries. ApsaraDB for HBase supports core services of Alibaba Group, such as Taobao, Alipay, and Cainiao.
Supported scenarios
ApsaraDB for HBase stores structured and semi-structured data that does not require strong transaction support. In addition to its NoSQL foundation, it supports SQL and secondary indexes.
The following table summarizes where ApsaraDB for HBase is most commonly deployed:
| Scenario | Core challenge | Representative use cases |
|---|---|---|
| Big data | Store and analyze massive datasets cost-effectively | Offline analytics with Spark and MaxCompute |
| Advertising | Handle real-time, high-concurrency targeting at low latency | Real-time bidding, targeted advertising |
| Internet of Vehicles (IoV) | Ingest high-frequency location and status data reliably | Ride-hailing, logistics, new energy vehicle detection |
| Social networking | Serve social feeds at scale with low latency | Feed management for messages, posts, and likes |
| Finance and retail | Process transactions and risk signals at high throughput | Transaction management, financial risk control |
Big data
Large-scale analytics pipelines face two competing pressures: storing massive datasets affordably while keeping query response times low enough for operational use. ApsaraDB for HBase resolves this tension with a decoupled storage and computing architecture that scales each layer independently.
Key capabilities for big data workloads:
Cost-effective storage: High compression ratios, cold and hot data separation, and tiered storage across hard disk drives (HDDs) and Object Storage Service (OSS) reduce storage costs at scale.
Efficient data ingestion: BulkLoad imports large datasets faster than row-by-row writes. Lindorm Tunnel Service (LTS) synchronizes full and incremental data from third-party systems with minimal setup.
Real-time query: Millisecond response times allow operational queries on datasets that would otherwise require a separate serving layer.
Offline analytics integration: Native integration with Spark and MaxCompute enables batch and offline analysis directly on data stored in ApsaraDB for HBase.
Horizontal scaling: Scale out to tens of millions of QPS without architectural changes.
Advertising
Advertising platforms require sub-millisecond reads for real-time bidding and the ability to handle massive traffic spikes — while storing constantly evolving user profiles, user behaviors, clickstreams, and advertising feeds. A rigid schema makes it difficult to add new features or targeting signals without downtime.
ApsaraDB for HBase addresses these requirements through:
Low latency: Millisecond response times for read-heavy workloads; master-replica dual clusters accelerate concurrent requests.
Flexible schema: Dynamic columns let you add or remove features and tags without schema migrations. Time to Live (TTL) settings automatically expire stale data.
High concurrency: Horizontal scaling supports tens of millions of QPS — sufficient for peak traffic during campaign launches.
Cost-effective storage: High compression ratios, cold and hot data separation, and HDD/OSS tiered storage lower the cost of retaining large volumes of historical behavioral data.
Data synchronization: LTS moves data between ApsaraDB for HBase and third-party ad tech systems efficiently.
High availability: Active-standby or active-active disaster recovery with automatic failover commits a 99.95% service level agreement (SLA).
Internet of Vehicles (IoV)
IoV systems continuously ingest driving trajectories, vehicle status, and precise positioning data from large fleets. The data volume is high, the write frequency is extreme, and query patterns range from real-time lookups to historical trajectory analysis. Storage costs compound quickly at fleet scale.
ApsaraDB for HBase handles these characteristics with:
Cost-effective storage: High compression ratios, cold and hot data separation, and HDD/OSS tiered storage make long-term retention of trajectory data economical.
Independent scalability: A decoupled storage and computing architecture lets you scale compute and storage separately as fleet size grows.
Flexible data model: Dynamic columns and TTL settings accommodate evolving telemetry schemas and automatic expiration of time-sensitive signals.
Data synchronization: LTS integrates ApsaraDB for HBase with third-party IoV platforms and data pipelines.
Low latency: Millisecond responses for real-time vehicle lookups; master-replica dual clusters support concurrent queries.
High availability: Active-standby or active-active disaster recovery with automatic failover commits a 99.95% SLA — critical for logistics and ride-hailing services where downtime has direct operational impact.
Social networking
Social feeds involve unpredictable traffic spikes, high write throughput for messages and interactions, and strict latency expectations on read paths. The data model must accommodate a wide variety of content types — messages, comments, posts, and likes — without sacrificing query performance.
ApsaraDB for HBase is well-suited to social networking workloads because it combines:
High availability: Active-standby or active-active disaster recovery with automatic failover commits a 99.95% SLA.
Low latency: Millisecond responses for read-heavy feed queries; master-replica dual clusters accelerate concurrent requests.
Flexible schema: Dynamic columns let you extend the data model as product features evolve, without schema migrations.
Horizontal scaling: Scale out to tens of millions of QPS to handle traffic surges without manual intervention.
Finance and retail
Financial and retail applications must handle high-throughput transaction writes, store large volumes of historical records, and power real-time risk control — all with strict latency and availability requirements. Risk control depends on fast access to user behaviors, user profiles, rule models, and fingerprint data to evaluate each transaction in milliseconds.
ApsaraDB for HBase meets these requirements through:
Cost-effective storage: High compression ratios, cold and hot data separation, and HDD/OSS tiered storage reduce the cost of retaining multi-year transaction histories.
High concurrency: Horizontal scaling supports tens of millions of QPS for peak transaction periods.
Flexible schema: Dynamic columns and TTL settings accommodate evolving risk models and automatic expiration of short-lived signals.
Low latency: Millisecond responses for risk scoring lookups; master-replica dual clusters accelerate concurrent requests.
Data synchronization: LTS moves data between ApsaraDB for HBase and core banking or retail systems efficiently.
High availability: Active-standby or active-active disaster recovery with automatic failover commits a 99.95% SLA — ensuring continuity for mission-critical transaction processing.