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 Lindorm. Compatible with open source Elasticsearch APIs, it seamlessly serves as an index store for the wide table engine to accelerate search queries. The search engine supports common business scenarios such as orders, bills, and logs, and provides vector search for AI scenarios as a foundational 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	Integrates with LindormTable as a search index for wide tables, providing unified multi-model storage and retrieval. Supports complex multi-dimensional queries with tokenization, fuzzy queries, aggregation and analysis, and sorting with pagination. Integrates with the vector engine and AI engines to provide hybrid full-text and vector retrieval for retrieval-augmented generation (RAG), improving the accuracy of large model responses.
Cloud-native elasticity	Uses a decoupled compute-storage architecture that supports on-demand scaling for compute and storage resources. Shard migrations complete in seconds without moving data, enabling fast 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 diagram shows the architecture of the Lindorm search engine:

As the core engine of Lindorm, the search engine provides a unified SQL interface for read and write operations. Key features include:

Intelligent index optimization: Automatically selects optimal indexes with SearchIndex technology to improve query efficiency.
Cloud-native architecture: Uses a decoupled storage and compute design. Partitioned data is stored persistently in the Lindorm Distributed File System (LDFS), enabling partition scheduling without data migration and elastic scaling in seconds.
Ecosystem compatibility: Natively compatible with Elasticsearch APIs and the ELK technology stack, making ES ecosystem components available out of the box.