wu-sheng/elasticsearch

Elasticsearch is a search engine that lets you store documents and then search through them quickly and flexibly. Think of it as a purpose-built database designed specifically for finding things by their content, whether that is social media posts, product catalogs, log files, or articles. You send it data in JSON format (a simple way to structure text), and it indexes that data so you can run fast searches against it later. At a high level, you interact with it over the web using standard HTTP requests. You can put data in by sending a document to a specific URL, retrieve that document by requesting it, and search across all your stored documents using either simple URL parameters or a JSON-based query language. The searches go beyond basic keyword matching: you can filter by date ranges, search within specific fields, and rank results with different priority levels. The project is built to run across multiple servers, which is what "distributed" means here. When you have a large dataset, it automatically splits your data into pieces called shards and copies them for safety. If one server goes down, your search keeps working because other servers hold copies of the data. You can start with one server on your laptop and add more as your needs grow, and the system rebalances itself. It also supports multiple separate indexes, so a single installation could serve different applications or different users with their own isolated data. A product team building an e-commerce site might use this to power product search with filters and relevance tuning. A platform with social content might store user posts and let people search across friends' content with higher ranking than strangers' content. It is also commonly used for searching through application logs at scale, since it handles large volumes of timestamped data well. One notable design choice is that it does not require you to define your data structure upfront. You can start dumping documents in and the system figures out field types automatically (like recognizing a date string), though you can customize the schema later if needed. It is built on top of Lucene, an established text-search library, and wraps that power in a simpler web-friendly interface. The project is open source under the Apache License.