ApsaraDB for HBase Performance-enhanced Edition is a fully managed, Apache HBase-compatible NoSQL database service built on the LindormTable engine. It delivers high throughput and millisecond latency at petabyte scale, with enterprise-grade reliability validated across Alibaba Group's core services since 2011.
What it is
ApsaraDB for HBase Performance-enhanced Edition is the commercialized form of Lindorm — the same engine that powers Taobao, Alipay, Cainiao, Youku, and AMAP inside Alibaba Group. The service is fully compatible with Apache HBase, so existing HBase clients and tooling work without modification.
Lindorm is a multi-model database service developed by Alibaba Cloud. Its wide table engine, LindormTable, is the kernel that powers ApsaraDB for HBase Performance-enhanced Edition. To access additional data models — including time series, search, and file storage — use Lindorm directly.
Core capabilities
| Capability | Description |
|---|---|
| High throughput | Processes hundreds of millions of requests per second (RPS), with tens of trillions of operations per day |
| Petabyte-scale storage | Stores hundreds of petabytes of semi-structured and unstructured data |
| Millisecond latency | Delivers consistent low-latency access under production workloads |
| Adjustable consistency | Configurable consistency levels to balance latency and durability requirements |
| Seamless scalability | Scales compute and storage independently without downtime |
| Multiple index types | Supports various index types for real-time data retrieval |
| High availability | Enterprise-grade fault tolerance validated during peak events such as the Double 11 Shopping Festival, the Spring Festival, and the Chinese National Day |
| Cost-effectiveness | Optimized cost structure for large-scale data storage and retrieval workloads |
| Apache HBase compatibility | Fully compatible with Apache HBase — migrate existing workloads without client changes |
Use cases
Big data storage: Store and retrieve large-scale semi-structured and unstructured data where unlimited scalability and high throughput are required.
Online services: Power latency-sensitive applications that require high availability and millisecond response times.
Multi-functional queries: Serve workloads that combine wide table scans, secondary index lookups, and real-time retrieval in a single service.
History of Lindorm
| Date | Milestone |
|---|---|
| March 2011 | Lindorm 1.0 released on Taobao, based on Apache HBase; used for Data Cube and log queue workloads |
| March 2012 | Scales to 100 TB of data and 100,000 RPS |
| May 2013 | Large-cluster support and multitenancy added |
| January 2014 | SQL support released, compatible with Phoenix |
| November 2015 | High availability architecture implemented; peak RPS exceeds 50 million |
| November 2016 | Server count reaches 10,000; adopted as the mid-tier storage platform for AMAP, Youku, and other Alibaba Group services |
| August 2017 | Apache HBase Community Edition commercialized on Alibaba Cloud and starts to provide services to enterprises |
| July 2018 | Lindorm 2.0 released with an optimized engine delivering 7x the performance of Apache HBase |
| August 2019 | ApsaraDB for HBase Performance-enhanced Edition commercialized with exclusive features for storage, indexing, and disaster recovery |
| June 2020 | Lindorm evolves into a cloud native multi-model database supporting wide table, time series, search, and file models |