[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M731VS3GA49V0JB7W04W7NFB":3,"skill-same:01M731VS3GA49V0JB7W04W7NFB":48},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":25,"commitSha":26,"skillDirectory":27,"forksCount":28,"repositoryHtmlUrl":29,"repositoryHomepageUrl":6,"ownerHtmlUrl":30,"ownerBlogUrl":31,"ownerDescription":32,"updatedAt":33,"install":34,"resources":47},"01M731VS3GA49V0JB7W04W7NFB","clickhouse-io",null,"ClickHouse数据库模式、查询优化、分析以及高性能分析工作负载的数据工程最佳实践。","SriramGandhiS\u002FSkills","SriramGandhiS","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F211849986?v=4","01M72WQCW12BYX71C6WN584TTA","Sriram S","01M72WQCWTJ2RD0ET1X31AF1RC",1,"MIT","https:\u002F\u002Fgithub.com\u002FSriramGandhiS\u002FSkills\u002Fblob\u002Fmaster\u002Fweb-assets\u002Fdocs\u002Fzh-CN\u002Fskills\u002Fclickhouse-io\u002FSKILL.md",[],[],"2026-11-11 15:11:14",false,"---\nname: clickhouse-io\ndescription: ClickHouse数据库模式、查询优化、分析以及高性能分析工作负载的数据工程最佳实践。\norigin: ECC\n---\n\n# ClickHouse 分析模式\n\n用于高性能分析和数据工程的 ClickHouse 特定模式。\n\n## 何时激活\n\n* 设计 ClickHouse 表架构（MergeTree 引擎选择）\n* 编写分析查询（聚合、窗口函数、连接）\n* 优化查询性能（分区裁剪、投影、物化视图）\n* 摄取大量数据（批量插入、Kafka 集成）\n* 为分析目的从 PostgreSQL\u002FMySQL 迁移到 ClickHouse\n* 实现实时仪表板或时间序列分析\n\n## 概述\n\nClickHouse 是一个用于在线分析处理 (OLAP) 的列式数据库管理系统 (DBMS)。它针对大型数据集上的快速分析查询进行了优化。\n\n**关键特性:**\n\n* 列式存储\n* 数据压缩\n* 并行查询执行\n* 分布式查询\n* 实时分析\n\n## 表设计模式\n\n### MergeTree 引擎 (最常用)\n\n```sql\nCREATE TABLE markets_analytics (\n    date Date,\n    market_id String,\n    market_name String,\n    volume UInt64,\n    trades UInt32,\n    unique_traders UInt32,\n    avg_trade_size Float64,\n    created_at DateTime\n) ENGINE = MergeTree()\nPARTITION BY toYYYYMM(date)\nORDER BY (date, market_id)\nSETTINGS index_granularity = 8192;\n```\n\n### ReplacingMergeTree (去重)\n\n```sql\n-- For data that may have duplicates (e.g., from multiple sources)\nCREATE TABLE user_events (\n    event_id String,\n    user_id String,\n    event_type String,\n    timestamp DateTime,\n    properties String\n) ENGINE = ReplacingMergeTree()\nPARTITION BY toYYYYMM(timestamp)\nORDER BY (user_id, event_id, timestamp)\nPRIMARY KEY (user_id, event_id);\n```\n\n### AggregatingMergeTree (预聚合)\n\n```sql\n-- For maintaining aggregated metrics\nCREATE TABLE market_stats_hourly (\n    hour DateTime,\n    market_id String,\n    total_volume AggregateFunction(sum, UInt64),\n    total_trades AggregateFunction(count, UInt32),\n    unique_users AggregateFunction(uniq, String)\n) ENGINE = AggregatingMergeTree()\nPARTITION BY toYYYYMM(hour)\nORDER BY (hour, market_id);\n\n-- Query aggregated data\nSELECT\n    hour,\n    market_id,\n    sumMerge(total_volume) AS volume,\n    countMerge(total_trades) AS trades,\n    uniqMerge(unique_users) AS users\nFROM market_stats_hourly\nWHERE hour >= toStartOfHour(now() - INTERVAL 24 HOUR)\nGROUP BY hour, market_id\nORDER BY hour DESC;\n```\n\n## 查询优化模式\n\n### 高效过滤\n\n```sql\n-- PASS: GOOD: Use indexed columns first\nSELECT *\nFROM markets_analytics\nWHERE date >= '2025-01-01'\n  AND market_id = 'market-123'\n  AND volume > 1000\nORDER BY date DESC\nLIMIT 100;\n\n-- FAIL: BAD: Filter on non-indexed columns first\nSELECT *\nFROM markets_analytics\nWHERE volume > 1000\n  AND market_name LIKE '%election%'\n  AND date >= '2025-01-01';\n```\n\n### 聚合\n\n```sql\n-- PASS: GOOD: Use ClickHouse-specific aggregation functions\nSELECT\n    toStartOfDay(created_at) AS day,\n    market_id,\n    sum(volume) AS total_volume,\n    count() AS total_trades,\n    uniq(trader_id) AS unique_traders,\n    avg(trade_size) AS avg_size\nFROM trades\nWHERE created_at >= today() - INTERVAL 7 DAY\nGROUP BY day, market_id\nORDER BY day DESC, total_volume DESC;\n\n-- PASS: Use quantile for percentiles (more efficient than percentile)\nSELECT\n    quantile(0.50)(trade_size) AS median,\n    quantile(0.95)(trade_size) AS p95,\n    quantile(0.99)(trade_size) AS p99\nFROM trades\nWHERE created_at >= now() - INTERVAL 1 HOUR;\n```\n\n### 窗口函数\n\n```sql\n-- Calculate running totals\nSELECT\n    date,\n    market_id,\n    volume,\n    sum(volume) OVER (\n        PARTITION BY market_id\n        ORDER BY date\n        ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW\n    ) AS cumulative_volume\nFROM markets_analytics\nWHERE date >= today() - INTERVAL 30 DAY\nORDER BY market_id, date;\n```\n\n## 数据插入模式\n\n### 批量插入 (推荐)\n\n```typescript\nimport { ClickHouse } from 'clickhouse'\n\nconst clickhouse = new ClickHouse({\n  url: process.env.CLICKHOUSE_URL,\n  port: 8123,\n  basicAuth: {\n    username: process.env.CLICKHOUSE_USER,\n    password: process.env.CLICKHOUSE_PASSWORD\n  }\n})\n\n\u002F\u002F PASS: Batch insert (efficient)\nasync function bulkInsertTrades(trades: Trade[]) {\n  const values = trades.map(trade => `(\n    '${trade.id}',\n    '${trade.market_id}',\n    '${trade.user_id}',\n    ${trade.amount},\n    '${trade.timestamp.toISOString()}'\n  )`).join(',')\n\n  await clickhouse.query(`\n    INSERT INTO trades (id, market_id, user_id, amount, timestamp)\n    VALUES ${values}\n  `).toPromise()\n}\n\n\u002F\u002F FAIL: Individual inserts (slow)\nasync function insertTrade(trade: Trade) {\n  \u002F\u002F Don't do this in a loop!\n  await clickhouse.query(`\n    INSERT INTO trades VALUES ('${trade.id}', ...)\n  `).toPromise()\n}\n```\n\n### 流式插入\n\n```typescript\n\u002F\u002F For continuous data ingestion\nimport { createWriteStream } from 'fs'\nimport { pipeline } from 'stream\u002Fpromises'\n\nasync function streamInserts() {\n  const stream = clickhouse.insert('trades').stream()\n\n  for await (const batch of dataSource) {\n    stream.write(batch)\n  }\n\n  await stream.end()\n}\n```\n\n## 物化视图\n\n### 实时聚合\n\n```sql\n-- Create materialized view for hourly stats\nCREATE MATERIALIZED VIEW market_stats_hourly_mv\nTO market_stats_hourly\nAS SELECT\n    toStartOfHour(timestamp) AS hour,\n    market_id,\n    sumState(amount) AS total_volume,\n    countState() AS total_trades,\n    uniqState(user_id) AS unique_users\nFROM trades\nGROUP BY hour, market_id;\n\n-- Query the materialized view\nSELECT\n    hour,\n    market_id,\n    sumMerge(total_volume) AS volume,\n    countMerge(total_trades) AS trades,\n    uniqMerge(unique_users) AS users\nFROM market_stats_hourly\nWHERE hour >= now() - INTERVAL 24 HOUR\nGROUP BY hour, market_id;\n```\n\n## 性能监控\n\n### 查询性能\n\n```sql\n-- Check slow queries\nSELECT\n    query_id,\n    user,\n    query,\n    query_duration_ms,\n    read_rows,\n    read_bytes,\n    memory_usage\nFROM system.query_log\nWHERE type = 'QueryFinish'\n  AND query_duration_ms > 1000\n  AND event_time >= now() - INTERVAL 1 HOUR\nORDER BY query_duration_ms DESC\nLIMIT 10;\n```\n\n### 表统计信息\n\n```sql\n-- Check table sizes\nSELECT\n    database,\n    table,\n    formatReadableSize(sum(bytes)) AS size,\n    sum(rows) AS rows,\n    max(modification_time) AS latest_modification\nFROM system.parts\nWHERE active\nGROUP BY database, table\nORDER BY sum(bytes) DESC;\n```\n\n## 常见分析查询\n\n### 时间序列分析\n\n```sql\n-- Daily active users\nSELECT\n    toDate(timestamp) AS date,\n    uniq(user_id) AS daily_active_users\nFROM events\nWHERE timestamp >= today() - INTERVAL 30 DAY\nGROUP BY date\nORDER BY date;\n\n-- Retention analysis\nSELECT\n    signup_date,\n    countIf(days_since_signup = 0) AS day_0,\n    countIf(days_since_signup = 1) AS day_1,\n    countIf(days_since_signup = 7) AS day_7,\n    countIf(days_since_signup = 30) AS day_30\nFROM (\n    SELECT\n        user_id,\n        min(toDate(timestamp)) AS signup_date,\n        toDate(timestamp) AS activity_date,\n        dateDiff('day', signup_date, activity_date) AS days_since_signup\n    FROM events\n    GROUP BY user_id, activity_date\n)\nGROUP BY signup_date\nORDER BY signup_date DESC;\n```\n\n### 漏斗分析\n\n```sql\n-- Conversion funnel\nSELECT\n    countIf(step = 'viewed_market') AS viewed,\n    countIf(step = 'clicked_trade') AS clicked,\n    countIf(step = 'completed_trade') AS completed,\n    round(clicked \u002F viewed * 100, 2) AS view_to_click_rate,\n    round(completed \u002F clicked * 100, 2) AS click_to_completion_rate\nFROM (\n    SELECT\n        user_id,\n        session_id,\n        event_type AS step\n    FROM events\n    WHERE event_date = today()\n)\nGROUP BY session_id;\n```\n\n### 队列分析\n\n```sql\n-- User cohorts by signup month\nSELECT\n    toStartOfMonth(signup_date) AS cohort,\n    toStartOfMonth(activity_date) AS month,\n    dateDiff('month', cohort, month) AS months_since_signup,\n    count(DISTINCT user_id) AS active_users\nFROM (\n    SELECT\n        user_id,\n        min(toDate(timestamp)) OVER (PARTITION BY user_id) AS signup_date,\n        toDate(timestamp) AS activity_date\n    FROM events\n)\nGROUP BY cohort, month, months_since_signup\nORDER BY cohort, months_since_signup;\n```\n\n## 数据流水线模式\n\n### ETL 模式\n\n```typescript\n\u002F\u002F Extract, Transform, Load\nasync function etlPipeline() {\n  \u002F\u002F 1. Extract from source\n  const rawData = await extractFromPostgres()\n\n  \u002F\u002F 2. Transform\n  const transformed = rawData.map(row => ({\n    date: new Date(row.created_at).toISOString().split('T')[0],\n    market_id: row.market_slug,\n    volume: parseFloat(row.total_volume),\n    trades: parseInt(row.trade_count)\n  }))\n\n  \u002F\u002F 3. Load to ClickHouse\n  await bulkInsertToClickHouse(transformed)\n}\n\n\u002F\u002F Run periodically\nsetInterval(etlPipeline, 60 * 60 * 1000)  \u002F\u002F Every hour\n```\n\n### 变更数据捕获 (CDC)\n\n```typescript\n\u002F\u002F Listen to PostgreSQL changes and sync to ClickHouse\nimport { Client } from 'pg'\n\nconst pgClient = new Client({ connectionString: process.env.DATABASE_URL })\n\npgClient.query('LISTEN market_updates')\n\npgClient.on('notification', async (msg) => {\n  const update = JSON.parse(msg.payload)\n\n  await clickhouse.insert('market_updates', [\n    {\n      market_id: update.id,\n      event_type: update.operation,  \u002F\u002F INSERT, UPDATE, DELETE\n      timestamp: new Date(),\n      data: JSON.stringify(update.new_data)\n    }\n  ])\n})\n```\n\n## 最佳实践\n\n### 1. 分区策略\n\n* 按时间分区 (通常是月或日)\n* 避免过多分区 (影响性能)\n* 对分区键使用 DATE 类型\n\n### 2. 排序键\n\n* 将最常过滤的列放在前面\n* 考虑基数 (高基数优先)\n* 排序影响压缩\n\n### 3. 数据类型\n\n* 使用最合适的较小类型 (UInt32 对比 UInt64)\n* 对重复字符串使用 LowCardinality\n* 对分类数据使用 Enum\n\n### 4. 避免\n\n* SELECT \\* (指定列)\n* FINAL (改为在查询前合并数据)\n* 过多的 JOIN (分析场景下进行反规范化)\n* 频繁的小批量插入 (改为批量)\n\n### 5. 监控\n\n* 跟踪查询性能\n* 监控磁盘使用情况\n* 检查合并操作\n* 查看慢查询日志\n\n**记住**: ClickHouse 擅长分析工作负载。根据查询模式设计表，批量插入，并利用物化视图进行实时聚合。\n","# ClickHouse 分析模式\n\n用于高性能分析和数据工程的 ClickHouse 特定模式。\n\n## 何时激活\n\n* 设计 ClickHouse 表架构（MergeTree 引擎选择）\n* 编写分析查询（聚合、窗口函数、连接）\n* 优化查询性能（分区裁剪、投影、物化视图）\n* 摄取大量数据（批量插入、Kafka 集成）\n* 为分析目的从 PostgreSQL\u002FMySQL 迁移到 ClickHouse\n* 实现实时仪表板或时间序列分析\n\n## 概述\n\nClickHouse 是一个用于在线分析处理 (OLAP) 的列式数据库管理系统 (DBMS)。它针对大型数据集上的快速分析查询进行了优化。\n\n**关键特性:**\n\n* 列式存储\n* 数据压缩\n* 并行查询执行\n* 分布式查询\n* 实时分析\n\n## 表设计模式\n\n### MergeTree 引擎 (最常用)\n\n```sql\nCREATE TABLE markets_analytics (\n    date Date,\n    market_id String,\n    market_name String,\n    volume UInt64,\n    trades UInt32,\n    unique_traders UInt32,\n    avg_trade_size Float64,\n    created_at DateTime\n) ENGINE = MergeTree()\nPARTITION BY toYYYYMM(date)\nORDER BY (date, market_id)\nSETTINGS index_granularity = 8192;\n```\n\n### ReplacingMergeTree (去重)\n\n```sql\n-- For data that may have duplicates (e.g., from multiple sources)\nCREATE TABLE user_events (\n    event_id String,\n    user_id String,\n    event_type String,\n    timestamp DateTime,\n    properties String\n) ENGINE = ReplacingMergeTree()\nPARTITION BY toYYYYMM(timestamp)\nORDER BY (user_id, event_id, timestamp)\nPRIMARY KEY (user_id, event_id);\n```\n\n### AggregatingMergeTree (预聚合)\n\n```sql\n-- For maintaining aggregated metrics\nCREATE TABLE market_stats_hourly (\n    hour DateTime,\n    market_id String,\n    total_volume AggregateFunction(sum, UInt64),\n    total_trades AggregateFunction(count, UInt32),\n    unique_users AggregateFunction(uniq, String)\n) ENGINE = AggregatingMergeTree()\nPARTITION BY toYYYYMM(hour)\nORDER BY (hour, market_id);\n\n-- Query aggregated data\nSELECT\n    hour,\n    market_id,\n    sumMerge(total_volume) AS volume,\n    countMerge(total_trades) AS trades,\n    uniqMerge(unique_users) AS users\nFROM market_stats_hourly\nWHERE hour >= toStartOfHour(now() - INTERVAL 24 HOUR)\nGROUP BY hour, market_id\nORDER BY hour DESC;\n```\n\n## 查询优化模式\n\n### 高效过滤\n\n```sql\n-- PASS: GOOD: Use indexed columns first\nSELECT *\nFROM markets_analytics\nWHERE date >= '2025-01-01'\n  AND market_id = 'market-123'\n  AND volume > 1000\nORDER BY date DESC\nLIMIT 100;\n\n-- FAIL: BAD: Filter on non-indexed columns first\nSELECT *\nFROM markets_analytics\nWHERE volume > 1000\n  AND market_name LIKE '%election%'\n  AND date >= '2025-01-01';\n```\n\n### 聚合\n\n```sql\n-- PASS: GOOD: Use ClickHouse-specific aggregation functions\nSELECT\n    toStartOfDay(created_at) AS day,\n    market_id,\n    sum(volume) AS total_volume,\n    count() AS total_trades,\n    uniq(trader_id) AS unique_traders,\n    avg(trade_size) AS avg_size\nFROM trades\nWHERE created_at >= today() - INTERVAL 7 DAY\nGROUP BY day, market_id\nORDER BY day DESC, total_volume DESC;\n\n-- PASS: Use quantile for percentiles (more efficient than percentile)\nSELECT\n    quantile(0.50)(trade_size) AS median,\n    quantile(0.95)(trade_size) AS p95,\n    quantile(0.99)(trade_size) AS p99\nFROM trades\nWHERE created_at >= now() - INTERVAL 1 HOUR;\n```\n\n### 窗口函数\n\n```sql\n-- Calculate running totals\nSELECT\n    date,\n    market_id,\n    volume,\n    sum(volume) OVER (\n        PARTITION BY market_id\n        ORDER BY date\n        ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW\n    ) AS cumulative_volume\nFROM markets_analytics\nWHERE date >= today() - INTERVAL 30 DAY\nORDER BY market_id, date;\n```\n\n## 数据插入模式\n\n### 批量插入 (推荐)\n\n```typescript\nimport { ClickHouse } from 'clickhouse'\n\nconst clickhouse = new ClickHouse({\n  url: process.env.CLICKHOUSE_URL,\n  port: 8123,\n  basicAuth: {\n    username: process.env.CLICKHOUSE_USER,\n    password: process.env.CLICKHOUSE_PASSWORD\n  }\n})\n\n\u002F\u002F PASS: Batch insert (efficient)\nasync function bulkInsertTrades(trades: Trade[]) {\n  const values = trades.map(trade => `(\n    '${trade.id}',\n    '${trade.market_id}',\n    '${trade.user_id}',\n    ${trade.amount},\n    '${trade.timestamp.toISOString()}'\n  )`).join(',')\n\n  await clickhouse.query(`\n    INSERT INTO trades (id, market_id, user_id, amount, timestamp)\n    VALUES ${values}\n  `).toPromise()\n}\n\n\u002F\u002F FAIL: Individual inserts (slow)\nasync function insertTrade(trade: Trade) {\n  \u002F\u002F Don't do this in a loop!\n  await clickhouse.query(`\n    INSERT INTO trades VALUES ('${trade.id}', ...)\n  `).toPromise()\n}\n```\n\n### 流式插入\n\n```typescript\n\u002F\u002F For continuous data ingestion\nimport { createWriteStream } from 'fs'\nimport { pipeline } from 'stream\u002Fpromises'\n\nasync function streamInserts() {\n  const stream = clickhouse.insert('trades').stream()\n\n  for await (const batch of dataSource) {\n    stream.write(batch)\n  }\n\n  await stream.end()\n}\n```\n\n## 物化视图\n\n### 实时聚合\n\n```sql\n-- Create materialized view for hourly stats\nCREATE MATERIALIZED VIEW market_stats_hourly_mv\nTO market_stats_hourly\nAS SELECT\n    toStartOfHour(timestamp) AS hour,\n    market_id,\n    sumState(amount) AS total_volume,\n    countState() AS total_trades,\n    uniqState(user_id) AS unique_users\nFROM trades\nGROUP BY hour, market_id;\n\n-- Query the materialized view\nSELECT\n    hour,\n    market_id,\n    sumMerge(total_volume) AS volume,\n    countMerge(total_trades) AS trades,\n    uniqMerge(unique_users) AS users\nFROM market_stats_hourly\nWHERE hour >= now() - INTERVAL 24 HOUR\nGROUP BY hour, market_id;\n```\n\n## 性能监控\n\n### 查询性能\n\n```sql\n-- Check slow queries\nSELECT\n    query_id,\n    user,\n    query,\n    query_duration_ms,\n    read_rows,\n    read_bytes,\n    memory_usage\nFROM system.query_log\nWHERE type = 'QueryFinish'\n  AND query_duration_ms > 1000\n  AND event_time >= now() - INTERVAL 1 HOUR\nORDER BY query_duration_ms DESC\nLIMIT 10;\n```\n\n### 表统计信息\n\n```sql\n-- Check table sizes\nSELECT\n    database,\n    table,\n    formatReadableSize(sum(bytes)) AS size,\n    sum(rows) AS rows,\n    max(modification_time) AS latest_modification\nFROM system.parts\nWHERE active\nGROUP BY database, table\nORDER BY sum(bytes) DESC;\n```\n\n## 常见分析查询\n\n### 时间序列分析\n\n```sql\n-- Daily active users\nSELECT\n    toDate(timestamp) AS date,\n    uniq(user_id) AS daily_active_users\nFROM events\nWHERE timestamp >= today() - INTERVAL 30 DAY\nGROUP BY date\nORDER BY date;\n\n-- Retention analysis\nSELECT\n    signup_date,\n    countIf(days_since_signup = 0) AS day_0,\n    countIf(days_since_signup = 1) AS day_1,\n    countIf(days_since_signup = 7) AS day_7,\n    countIf(days_since_signup = 30) AS day_30\nFROM (\n    SELECT\n        user_id,\n        min(toDate(timestamp)) AS signup_date,\n        toDate(timestamp) AS activity_date,\n        dateDiff('day', signup_date, activity_date) AS days_since_signup\n    FROM events\n    GROUP BY user_id, activity_date\n)\nGROUP BY signup_date\nORDER BY signup_date DESC;\n```\n\n### 漏斗分析\n\n```sql\n-- Conversion funnel\nSELECT\n    countIf(step = 'viewed_market') AS viewed,\n    countIf(step = 'clicked_trade') AS clicked,\n    countIf(step = 'completed_trade') AS completed,\n    round(clicked \u002F viewed * 100, 2) AS view_to_click_rate,\n    round(completed \u002F clicked * 100, 2) AS click_to_completion_rate\nFROM (\n    SELECT\n        user_id,\n        session_id,\n        event_type AS step\n    FROM events\n    WHERE event_date = today()\n)\nGROUP BY session_id;\n```\n\n### 队列分析\n\n```sql\n-- User cohorts by signup month\nSELECT\n    toStartOfMonth(signup_date) AS cohort,\n    toStartOfMonth(activity_date) AS month,\n    dateDiff('month', cohort, month) AS months_since_signup,\n    count(DISTINCT user_id) AS active_users\nFROM (\n    SELECT\n        user_id,\n        min(toDate(timestamp)) OVER (PARTITION BY user_id) AS signup_date,\n        toDate(timestamp) AS activity_date\n    FROM events\n)\nGROUP BY cohort, month, months_since_signup\nORDER BY cohort, months_since_signup;\n```\n\n## 数据流水线模式\n\n### ETL 模式\n\n```typescript\n\u002F\u002F Extract, Transform, Load\nasync function etlPipeline() {\n  \u002F\u002F 1. Extract from source\n  const rawData = await extractFromPostgres()\n\n  \u002F\u002F 2. Transform\n  const transformed = rawData.map(row => ({\n    date: new Date(row.created_at).toISOString().split('T')[0],\n    market_id: row.market_slug,\n    volume: parseFloat(row.total_volume),\n    trades: parseInt(row.trade_count)\n  }))\n\n  \u002F\u002F 3. Load to ClickHouse\n  await bulkInsertToClickHouse(transformed)\n}\n\n\u002F\u002F Run periodically\nsetInterval(etlPipeline, 60 * 60 * 1000)  \u002F\u002F Every hour\n```\n\n### 变更数据捕获 (CDC)\n\n```typescript\n\u002F\u002F Listen to PostgreSQL changes and sync to ClickHouse\nimport { Client } from 'pg'\n\nconst pgClient = new Client({ connectionString: process.env.DATABASE_URL })\n\npgClient.query('LISTEN market_updates')\n\npgClient.on('notification', async (msg) => {\n  const update = JSON.parse(msg.payload)\n\n  await clickhouse.insert('market_updates', [\n    {\n      market_id: update.id,\n      event_type: update.operation,  \u002F\u002F INSERT, UPDATE, DELETE\n      timestamp: new Date(),\n      data: JSON.stringify(update.new_data)\n    }\n  ])\n})\n```\n\n## 最佳实践\n\n### 1. 分区策略\n\n* 按时间分区 (通常是月或日)\n* 避免过多分区 (影响性能)\n* 对分区键使用 DATE 类型\n\n### 2. 排序键\n\n* 将最常过滤的列放在前面\n* 考虑基数 (高基数优先)\n* 排序影响压缩\n\n### 3. 数据类型\n\n* 使用最合适的较小类型 (UInt32 对比 UInt64)\n* 对重复字符串使用 LowCardinality\n* 对分类数据使用 Enum\n\n### 4. 避免\n\n* SELECT \\* (指定列)\n* FINAL (改为在查询前合并数据)\n* 过多的 JOIN (分析场景下进行反规范化)\n* 频繁的小批量插入 (改为批量)\n\n### 5. 监控\n\n* 跟踪查询性能\n* 监控磁盘使用情况\n* 检查合并操作\n* 查看慢查询日志\n\n**记住**: ClickHouse 擅长分析工作负载。根据查询模式设计表，批量插入，并利用物化视图进行实时聚合。\n",{"name":5,"origin":24,"description":7},"ECC","master","d6ca01da0f24a3b0e26e4a8b7b66cbd25d429f5d","web-assets\u002Fdocs\u002Fzh-CN\u002Fskills\u002Fclickhouse-io",0,"https:\u002F\u002Fgithub.com\u002FSriramGandhiS\u002FSkills","https:\u002F\u002Fgithub.com\u002FSriramGandhiS","https:\u002F\u002Fsriram.website\u002F","Curious guy who keeps tinkering until something works","2026-05-20 16:21:34",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":46},[36,39,42],{"tool":37,"command":38},"npx","npx skills add SriramGandhiS\u002FSkills --skill clickhouse-io",{"tool":40,"command":41},"bunx","bunx skills add SriramGandhiS\u002FSkills --skill clickhouse-io",{"tool":43,"command":44},"pnpm","pnpm dlx skills add SriramGandhiS\u002FSkills --skill clickhouse-io","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FSriramGandhiS\u002FSkills\u002Fblob\u002Fmaster\u002Fweb-assets\u002Fdocs\u002Fzh-CN\u002Fskills\u002Fclickhouse-io\u002FSKILL.md",true,[],[49,56,61,67,74,81],{"skillNo":50,"name":51,"nameZh":6,"description":52,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":53,"categories":54,"keywords":55,"publishedAt":19},"01M732609EPJWV13PVTX3GT4TM","webgpu-threejs-tsl","Comprehensive guide for developing WebGPU-enabled Three.js applications using TSL (Three.js Shading Language). Covers WebGPU renderer setup, TSL syntax and node materials, compute shaders, post-processing effects, and WGSL integration. Use this skill when working with Three.js WebGPU, TSL shaders, node materials, or GPU compute in Three.js.","https:\u002F\u002Fgithub.com\u002FSriramGandhiS\u002FSkills\u002Fblob\u002Fmaster\u002Fwebgpu-threejs-tsl\u002Fskills\u002Fwebgpu-threejs-tsl\u002FSKILL.md",[],[],{"skillNo":57,"name":51,"nameZh":6,"description":52,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":58,"categories":59,"keywords":60,"publishedAt":19},"01M7325XM47AGWB6K6C4VXKE6B","https:\u002F\u002Fgithub.com\u002FSriramGandhiS\u002FSkills\u002Fblob\u002Fmaster\u002Fwebgpu-threejs-tsl\u002FSKILL.md",[],[],{"skillNo":62,"name":63,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":64,"categories":65,"keywords":66,"publishedAt":19},"01M7325WP6DES69JKC840VMFEP","verification-loop","https:\u002F\u002Fgithub.com\u002FSriramGandhiS\u002FSkills\u002Fblob\u002Fmaster\u002Fweb-assets\u002Fdocs\u002Fzh-TW\u002Fskills\u002Fverification-loop\u002FSKILL.md",[],[],{"skillNo":68,"name":69,"nameZh":6,"description":70,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":71,"categories":72,"keywords":73,"publishedAt":19},"01M7325TFQF1DZTQ6J3WGQQ1VR","tdd-workflow","Use this skill when writing new features, fixing bugs, or refactoring code. Enforces test-driven development with 80%+ coverage including unit, integration, and E2E tests.","https:\u002F\u002Fgithub.com\u002FSriramGandhiS\u002FSkills\u002Fblob\u002Fmaster\u002Fweb-assets\u002Fdocs\u002Fzh-TW\u002Fskills\u002Ftdd-workflow\u002FSKILL.md",[],[],{"skillNo":75,"name":76,"nameZh":6,"description":77,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":78,"categories":79,"keywords":80,"publishedAt":19},"01M7325RGV9AY9PK92D2V8KKPW","strategic-compact","Suggests manual context compaction at logical intervals to preserve context through task phases rather than arbitrary auto-compaction.","https:\u002F\u002Fgithub.com\u002FSriramGandhiS\u002FSkills\u002Fblob\u002Fmaster\u002Fweb-assets\u002Fdocs\u002Fzh-TW\u002Fskills\u002Fstrategic-compact\u002FSKILL.md",[],[],{"skillNo":82,"name":83,"nameZh":6,"description":84,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":85,"categories":86,"keywords":87,"publishedAt":19},"01M7325MZC28SE86E7NMCNEFS4","security-review","Use this skill when adding authentication, handling user input, working with secrets, creating API endpoints, or implementing payment\u002Fsensitive features. Provides comprehensive security checklist and patterns.","https:\u002F\u002Fgithub.com\u002FSriramGandhiS\u002FSkills\u002Fblob\u002Fmaster\u002Fweb-assets\u002Fdocs\u002Fzh-TW\u002Fskills\u002Fsecurity-review\u002FSKILL.md",[],[]]