ApsaraDB for HBase Performance-enhanced Edition (Lindorm) is a fully managed cloud service that extends Apache HBase with enterprise-grade performance, reliability, and ecosystem integrations — without the operational overhead of managing clusters yourself.
Core features
Lindorm supports the HBase API alongside additional data models and built-in indexing that Apache HBase does not provide out of the box.
| Feature | Lindorm | Apache HBase |
|---|---|---|
| HBase API | Supported | Supported |
| Data models | Wide columns (HBase API), tables (SQL-like API), queues, and more | Wide columns only |
| Global secondary index (GSI) | Built-in; transparent queries with configurable redundancy for non-index key columns. See Secondary indexes. | Requires external components; complex to configure |
| Full-text search | Integrated with Apache Solr; unified storage, multi-dimensional queries, and full-text indexes across massive datasets. See Full-text index service. | Not supported |
Performance
Lindorm delivers significantly higher throughput and lower latency than Apache HBase through deep engine-level optimizations. These gains translate directly to higher concurrency headroom and more consistent response times during peak traffic.
| Metric | Lindorm | Apache HBase |
|---|---|---|
| Throughput | 7x that of Apache HBase. See Test results. | Baseline |
| P99 latency | 1/10 of Apache HBase. See Test results. | Baseline |
Cost efficiency
Lindorm reduces storage costs through better compression algorithms, smarter encoding, and automatic cold/hot data tiering — without requiring manual tuning.
| Feature | Lindorm | Apache HBase |
|---|---|---|
| Data compression | Uses a deeply optimized Zstandard (ZSTD) algorithm rewritten with JNA to eliminate core dumps, and optimized with dictionary sampling. The compression ratio is 50% higher than Snappy, with a data compression ratio of up to 10:1. | Uses Snappy by default. Zstandard requires a Hadoop-3.0 environment and may cause core dumps. |
| Encoding | Supports the IndexableDelta algorithm, which matches the DIFF algorithm's compression ratio while delivering 2x faster random access. | Uses DIFF by default, which has slower random access. |
| Cold and hot data separation | Automatically stores cold data in high-compression, low-cost media using tiered storage. This reduces storage costs by 70% and improves hot data query performance by 15%. See Cold and hot data separation. | Not supported |
| Storage media | Ultra disks, standard SSDs, local HDDs, local SSDs, Object Storage Service (OSS) for cold data, and capacity-optimized disks (coming soon) | N/A |
Reliability
Lindorm provides enterprise-grade reliability features including automated failover, large-scale backup, and fast fault recovery — backed by a service level agreement (SLA).
| Feature | Lindorm | Apache HBase |
|---|---|---|
| Active-standby redundancy | Supports automatic failover and dual-cluster deployment for concurrent request processing. Can run in active/standby mode alongside a self-managed Apache HBase instance. | No automated failover |
| Backup and restoration | Back up over 100 TB to OSS with a recovery time objective (RTO) of under 30 minutes, regardless of data size. Supports on-demand backup and point-in-time restoration. See Enable data backup and restoration. | Not supported |
| Mean time to repair (MTTR) | Fault recovery is 10x faster than Apache HBase. | Not supported |
Multitenancy
Lindorm supports secure multi-tenant deployments with fine-grained access control and physical resource isolation between tenants.
| Feature | Lindorm | Apache HBase |
|---|---|---|
| Authentication and ACL | Username/password authentication with access control list (ACL)-based permission management. See Manage users and ACLs. | Complex authentication process |
| Resource isolation | Resource group feature for physical isolation between tenants | Not supported |
O&M and diagnostics
Lindorm provides GUI-based tools for cluster management and data querying, replacing HBase Shell as the primary management interface.
| Feature | Lindorm | Apache HBase |
|---|---|---|
| O&M tools | GUI-based cluster management for tables, namespaces, groups, and ACLs. See Cluster management system. | HBase Shell only |
| Data query | HBase Shell and GUI-based SQL query in the cluster management system. See Data query. | HBase Shell only |
Ecosystem integrations
Lindorm integrates with major data services for migration, synchronization, and analytics — removing the heavy development work that self-managed HBase requires.
| Integration | Lindorm | Apache HBase |
|---|---|---|
| Data migration | Efficient, automated, online migration between HBase versions with no application impact and no code changes required. See Introduction to LTS (formerly known as BDS). | Offline migration only |
| MySQL synchronization | Uses Lindorm Tunnel Service (LTS) for full data sync and real-time online incremental sync from MySQL. See Introduction to LTS (formerly known as BDS). | No dedicated tools; no online incremental sync; requires third-party tools |
| Apache Spark analytics | Deeply integrated with Apache Spark: run Spark SQL queries against Lindorm data, archive incremental data to Spark, HDFS, or OSS, and write offline analysis results back to Lindorm. | No built-in integration; heavy development workloads required |
| MaxCompute | Export full data to MaxCompute. See Export full data to MaxCompute. | Heavy development workloads required |
| Log Service | Use LTS to import incremental data from Log Service. See Introduction to LTS (formerly known as BDS). | Heavy development workloads required |
Service and support
Lindorm is a fully managed service backed by an SLA and an expert team with deep Apache HBase experience.
| Aspect | Details |
|---|---|
| SLA | Up to 99.9% service availability for single-cluster deployment and 99.99% for dual-cluster deployment |
| O&M | Fully managed — no complex database operations required |
| Technical team | Supported by a team that includes Apache Project Management Committee (PMC) members and committers |
| Proven at scale | Nine years of continuous service during the Tmall Double 11 Shopping Festival, with tens of thousands of instances supporting Alibaba Group businesses |