Lindorm works alongside ApsaraDB RDS to offload historical data from relational databases into a low-cost, auto-scaling archive tier—cutting storage costs by more than 90% while preserving SQL query access and real-time read performance.
Only Lindorm Tunnel Service (LTS) instances purchased before March 10, 2023 can be used in this solution.
The problem: relational databases weren't built for archival scale
In mobile internet, e-commerce, and fintech workloads, daily data volumes grow exponentially while query frequency on older records drops over time. Keeping everything in a relational database creates three compounding problems:
-
Storage costs scale with data volume. Exponential data growth leads to exponential cost growth.
-
Query performance degrades above 1 TB. Once an instance exceeds 1 TB of storage, query latency increases noticeably.
-
Sharding multiplies O&M complexity. Database and table sharding relieves performance pressure but drives up development and operational costs.
How it works
The solution replaces a single overloaded relational database with a two-tier architecture:
Key capabilities
Cost reduction
Hot/cold data separation. Lindorm automatically classifies data as hot or cold based on access patterns. Hot data stays on high-performance storage; cold data moves to low-cost storage. The price ratio between the two tiers can reach 10:1, and no application code changes are required to read or write tables after separation.
Adaptive compression. Lindorm selects an encoding algorithm—dictionary, prefix, delta, or entropy encoding—based on the data type. Compared with standard industry algorithms, this improves compression ratios by 10%–30%.
Capacity-optimized disks with a built-in buffer layer. The buffer layer accelerates reads so that archived data queries return at latencies comparable to online queries.
Costs as low as0.11yuan/GB/month，Automatic scaling
Decoupled compute and storage. Compute and storage scale independently, so you right-size each dimension without overprovisioning the other.
Lindorm Serverless. Scales resources automatically based on traffic, uses pay-as-you-go billing, and eliminates manual capacity planning. Built on multi-tenant isolation and an Infrastructure as a Service (IaaS) platform, Lindorm Serverless provides a service level agreement (SLA) suited to enterprise availability requirements.
Query compatibility
Open source API compatibility. Lindorm is compatible with the Apache HBase API, Apache Phoenix API (SQL), and Cassandra API (Cassandra Query Language, CQL). Existing application code that uses these interfaces requires no modification after migration.
Rich query features. Global secondary indexes, multi-dimensional queries, dynamic columns, and time-to-live (TTL) settings are supported, covering use cases such as orders, bills, chat records, feed streams, images, and logs.
LindormSearch. An integrated search engine compatible with the Apache Solr API. Activate it on any Lindorm instance to enable full-text search, value aggregation, and complex multi-dimensional queries without external search infrastructure.
LindormTable benchmark performance. LindormTable throughput and latency benchmark at 7x the performance of open source Apache HBase. See Test results for details.
LindormTSDB benchmark performance. The benchmark performance of LindormTSDB is ranked first in the list released by the China Academy of Information and Communications Technology.
Monitoring and operations
Lindorm provides a comprehensive monitoring and alerting feature to help you ensure the stability of data migration tasks.
Big data integration
Connect Lindorm directly to Apache Spark, Apache Hive, Apache Flink, or Presto for batch analytics, streaming pipelines, and ad-hoc queries over the full historical dataset.
Use cases
