Lindorm search engine (compatible with Elasticsearch API)-Lindorm(Lindorm)-阿里云帮助中心

Lindorm search engine is a high-performance, cost-effective, and reliable distributed search engine built on the core of Lindorm. It is compatible with open source Elasticsearch APIs and can be seamlessly used as an index store for the wide table engine to accelerate search queries. The search engine is suitable for mainstream business scenarios, such as orders, bills, and logs. It also supports vector search for AI scenarios and is a fundamental service component for building large model applications.

Features

Core capability	Description
Ecosystem compatibility	Compatible with standard open source Elasticsearch APIs. Supports integration with the open source ELKB ecosystem.
Unified Multimodal Integration	Seamlessly integrates with LindormTable to act as a search index for wide tables. This provides unified multi-model storage and retrieval. The search index is designed for complex multi-dimensional query scenarios and supports features such as tokenization, fuzzy queries, aggregation and analysis, and sorting with pagination. Integrates with the vector engine and AI engines to provide the hybrid full-text and vector retrieval capabilities required by retrieval-augmented generation (RAG) systems. This improves the accuracy of large model responses.
Cloud-native elasticity	Features a decoupled compute and storage architecture. This supports on-demand scaling for compute and storage resources. Shard migrations complete in seconds without moving data, enabling fast and efficient node scaling.
Low cost	Supports multiple low-cost storage media, hot and cold data separation, and high-compression algorithms. This reduces storage costs by 50% compared to self-managed deployments. Supports enterprise-grade time to live (TTL) for defining data lifecycles at the index level. Expired data is automatically reclaimed to reduce costs.

Service architecture

The following figure shows the architecture of the Lindorm search engine:

The search engine is the core engine of Lindorm. It provides a unified SQL interface for read and write operations. Its features include the following:

Intelligent index optimization: It automatically selects optimal indexes using SearchIndex technology to significantly improve query efficiency.
Cloud-native architecture: It uses a decoupled storage and compute design. Partitioned data is persistently stored in the Lindorm Distributed File System (LDFS). This enables partition scheduling without data migration and supports elastic scaling in seconds.
Ecosystem compatibility: It is natively compatible with Elasticsearch APIs and the ELK technology stack. This ensures that ES ecosystem components are available out-of-the-box.