duckdb导出Excel和导出CSV速度测试

运行duckdb数据库

D:>duckdb
v1.2.0 5f5512b827
Enter “.help” for usage hints.
Connected to a transient in-memory database.
Use “.open FILENAME” to reopen on a persistent database.

生成模拟数据，10个列，100万行数据；

-- 步骤1：创建生成模拟数据的视图
CREATE OR REPLACE VIEW test_data_view AS
SELECT 
  -- 生成自增ID
  CAST(generate_series AS INTEGER) AS id,
  
  -- 随机字符串（3个大写字母）
  (
    CHR((RANDOM() * 26 + 65)::INTEGER) || 
    CHR((RANDOM() * 26 + 65)::INTEGER) || 
    CHR((RANDOM() * 26 + 65)::INTEGER)
  ) AS code,
  
  -- 可变长度文本（5-10字符）
  SUBSTRING(
    REPLACE(RANDOM()::STRING, '0.', '') || 
    REPLACE(RANDOM()::STRING, '0.', ''),
    1, 5 + (RANDOM() * 5)::INTEGER
  ) AS dynamic_text,
  
  -- 数值类型
  ROUND(RANDOM() * 1000, 2) AS price,         -- 带2位小数的数值
  (RANDOM() * 10000)::INTEGER AS quantity,    -- 整数库存量
  RANDOM()::BOOLEAN AS is_available,          -- 布尔值
  
  -- 日期时间类型
  DATE '2024-01-01' AS fixed_date,            -- 固定日期
  CURRENT_TIMESTAMP AT TIME ZONE 'UTC' AS utc_created,  -- 当前UTC时间
  
  -- 组合类型
  MAP(ARRAY['A', 'B', 'C'], 
      ARRAY[RANDOM(), RANDOM(), RANDOM()]) AS attributes,
  
  -- 生成范围约束数值（0-100）
  LEAST(GREATEST(RANDOM() * 120, 0), 100) AS constrained_value
  
FROM GENERATE_SERIES(1, 10000000);  -- 生成100万行测试数据

查看一下视图，发现实时随机生成100万行数据要3秒

所以又建一个t1表，复制这100万行数据

CREATE table t1 as select * from main.test_data_view;

基于T1表进行测试excel导出， 这样估计是最快的

1万行，用时0.4秒

5万行，用时2秒

10万行，用时4秒

20万行，用时8秒

100万行，用时35秒

D INSTALL excel;
100% ▕████████████████████████████████████████████████████████████▏
Run Time (s): real 33.381 user 13.625000 sys 0.093750
D LOAD excel;
Run Time (s): real 0.077 user 0.015625 sys 0.000000
D COPY t1 TO 'D:\100w.xlsx' WITH (FORMAT XLSX, HEADER TRUE);
100% ▕████████████████████████████████████████████████████████████▏
Run Time (s): real 34.868 user 26.968750 sys 0.062500
D COPY (SELECT * FROM t1 limit 10000 ) TO 'D:\test1w.xlsx' WITH (FORMAT XLSX, HEADER TRUE);
Run Time (s): real 0.386 user 0.640625 sys 0.000000
D COPY (SELECT * FROM t1 limit 50000 ) TO 'D:\test5w.xlsx' WITH (FORMAT XLSX, HEADER TRUE);
Run Time (s): real 1.825 user 1.390625 sys 0.015625
D COPY (SELECT * FROM t1 limit 100000 ) TO 'D:\test10w.xlsx' WITH (FORMAT XLSX, HEADER TRUE);
100% ▕████████████████████████████████████████████████████████████▏
Run Time (s): real 3.793 user 2.875000 sys 0.031250
D COPY (SELECT * FROM t1 limit 200000 ) TO 'D:\test20w.xlsx' WITH (FORMAT XLSX, HEADER TRUE);
100% ▕████████████████████████████████████████████████████████████▏
Run Time (s): real 7.815 user 12.984375 sys 0.031250
D COPY (SELECT * FROM t1 limit 500000 ) TO 'D:\test50w.xlsx' WITH (FORMAT XLSX, HEADER TRUE);
100% ▕████████████████████████████████████████████████████████████▏
Run Time (s): real 18.385 user 14.921875 sys 0.062500
D COPY (SELECT * FROM t1 limit 1000000 ) TO 'D:\test100w.xlsx' WITH (FORMAT XLSX, HEADER TRUE);
100% ▕████████████████████████████████████████████████████████████▏
Run Time (s): real 36.913 user 58.718750 sys 0.218750
D COPY t1 TO 'D:\test100w_2.xlsx' WITH (FORMAT XLSX, HEADER TRUE);
100% ▕████████████████████████████████████████████████████████████▏
Run Time (s): real 34.845 user 29.859375 sys 0.062500

导出CSV速度

1万行，0.05秒

10万行，0.4秒

50万行，1.2秒

100万行，3.2秒

D COPY (SELECT * FROM t1 limit   10000) TO 'D:\test1w.csv' (FORMAT CSV, HEADER);
Run Time (s): real 0.046 user 0.015625 sys 0.000000
D COPY (SELECT * FROM t1 limit  100000) TO 'D:\test10w.csv' (FORMAT CSV, HEADER);
Run Time (s): real 0.382 user 0.265625 sys 0.000000
D COPY (SELECT * FROM t1 limit  500000) TO 'D:\test50w.csv' (FORMAT CSV, HEADER);
Run Time (s): real 1.183 user 1.187500 sys 0.015625
D COPY (SELECT * FROM t1 limit 1000000) TO 'D:\test100w.csv' (FORMAT CSV, HEADER);
100% ▕████████████████████████████████████████████████████████████▏
Run Time (s): real 3.156 user 2.015625 sys 0.312500

最后也证明了猜测，从视图实时生成100万随机数据，再导出Excel差3秒，导出csv只差1.5秒

D select count(*) from test_data_view;
┌────────────────┐
│  count_star()  │
│     int64      │
├────────────────┤
│    1000000     │
│ (1.00 million) │
└────────────────┘
Run Time (s): real 0.007 user 0.000000 sys 0.000000
D COPY test_data_view TO 'D:\test100w_view.xlsx' WITH (FORMAT XLSX, HEADER TRUE);
100% ▕████████████████████████████████████████████████████████████▏
Run Time (s): real 37.115 user 27.640625 sys 0.093750
D COPY test_data_view TO 'D:\test100w_view.csv' (FORMAT CSV, HEADER);
100% ▕████████████████████████████████████████████████████████████▏
Run Time (s): real 4.641 user 3.656250 sys 0.140625