Databend 产品月报（2025 年 5 月）

-- 创建应用日志表CREATE OR REPLACE TABLE app_logs (id INT,timestamp TIMESTAMP,level STRING,message STRING);
-- 创建 ngram index（按3个字符分段）CREATE OR REPLACE NGRAM INDEX logs_message_idx ON app_logs(message) gram_size = 3;
-- 插入数据，index 会自动更新INSERT INTO app_logs VALUES (1, '2025-05-01 10:00:00', 'ERROR', 'Database connection failed');INSERT INTO app_logs VALUES (2, '2025-05-01 10:01:00', 'INFO', 'User login successful');INSERT INTO app_logs VALUES (3, '2025-05-01 10:02:00', 'ERROR', 'Kubernetes pod restart failed');
-- 原来的 LIKE 查询现在飞快SELECT * FROM app_logs WHERE message LIKE '%connection%';SELECT * FROM app_logs WHERE message LIKE '%kubernetes%';
-- 看看 index 是否生效EXPLAIN SELECT * FROM app_logs WHERE message LIKE '%error%';-- 会显示：bloom pruning: X to Y（说明 index 在工作）

-- 创建知识库表CREATE OR REPLACE TABLE knowledge_base (doc_id INT,title STRING,content STRING,category STRING);
-- 创建全文搜索 indexCREATE OR REPLACE INVERTED INDEX kb_content_idx ON knowledge_base(title, content, category);
-- 插入文档数据，index 自动维护INSERT INTO knowledge_base VALUES(1, 'Database Setup', 'Learn how to configure database connections and troubleshoot common issues', 'tutorial');
INSERT INTO knowledge_base VALUES (2, 'Kubernetes Guide', 'Complete guide to deploying applications on Kubernetes clusters', 'guide');
INSERT INTO knowledge_base VALUES(3, 'Error Handling', 'Best practices for error handling and logging in distributed systems', 'best-practices');
-- 基础搜索，带相关性打分（Elasticsearch 用户很熟悉）SELECT doc_id, title, SCORE() as relevanceFROM knowledge_baseWHERE MATCH(content, 'database troubleshoot')ORDER BY SCORE() DESC;
-- 模糊搜索 - 和 Elasticsearch 一模一样的语法SELECT * FROM knowledge_baseWHERE MATCH(content, 'kuberntes', 'fuzziness=1');
-- 高级查询，完全是 Elasticsearch 的写法SELECT * FROM knowledge_baseWHERE QUERY('content:"error handling" AND category:best-practices');
-- 布尔查询（Elasticsearch 用户几乎无学习成本）SELECT * FROM knowledge_baseWHERE QUERY('title:database OR kubernetes AND content:guide');