OceanBase SeekDB SQL优化案例---MySQL在客户端会没有市场的
A
liuaustin3约 23 分钟
liuaustin3约 23 分钟❝开头还是介绍一下群，如果感兴趣PolarDB ,MongoDB ,MySQL ,PostgreSQL ,Redis, OceanBase, Sql Server等有问题，有需求都可以加群群内有各大数据库行业大咖，可以解决你的问题。加群请联系 liuaustin3 ，（共3300人左右 1 + 2 + 3 + 4 +5 + 6 + 7 + 8 +9）(1 2 3 4 5 6 7群均已爆满，开8群近400 9群 200+，开10群PolarDB专业学习群100+)
上一篇帖子关于SEEKDB在线下终端市场的布局，做了一个预期，那么我们必须针对线下单机市场中的对于SEEKDB 在一些复杂SQL的运算进行测试，看看SEEKDB可以充当替换MYSQL线下单机市场的国产首选吗？ （一些同学看我上一篇帖子，说要试试SEEKDB）
测试会建立四张表，然后针对这些表进行复杂的查询并查看相关的执行计划，进行分析。
[root@seekdb ~]# obclient -h 127.0.0.1 -P 2881 -u root@sys
Welcome to the OceanBase. Commands end with ; or \g.
Your OceanBase connection id is 3221488705
Server version: OceanBase 4.3.5.3 SeekDB (r1.0.0.0) (Built 100000262025111218-5343637512e28c346f938516af53b7879d4d5974 Nov 12 2025)
Copyright (c) 2000, 2018, OceanBase and/or its affiliates. All rights reserved.
Type 'help;' or '\h'forhelp. Type '\c' to clear the current input statement.
obclient(root@sys)[(none)]> use test
Database changed
obclient(root@sys)[test]> use ob_test
ERROR 1049 (42000): Unknown database
obclient(root@sys)[test]> create database ob_test;
Query OK, 1 row affected (0.062 sec)
obclient(root@sys)[test]> use ob_test;
Database changed
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> CREATE TABLE customers (
-> customer_id BIGINT PRIMARY KEY,
-> region VARCHAR(32),
-> level INT,
-> created_at DATE
-> );
Query OK, 0 rows affected (0.144 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> CREATE INDEX idx_customers_region_level
-> ON customers(region, level);
Query OK, 0 rows affected (0.643 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> CREATE TABLE products (
-> product_id BIGINT PRIMARY KEY,
-> category VARCHAR(32),
-> price DECIMAL(10,2),
-> status CHAR(1)
-> );
Query OK, 0 rows affected (0.059 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> CREATE INDEX idx_products_category_status
-> ON products(category, status);
Query OK, 0 rows affected (0.494 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> CREATE TABLE orders (
-> order_id BIGINT PRIMARY KEY,
-> customer_id BIGINT,
-> order_date DATE,
-> order_status CHAR(1),
-> total_amount DECIMAL(12,2)
-> );
Query OK, 0 rows affected (0.072 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> CREATE INDEX idx_orders_customer_date
-> ON orders(customer_id, order_date);
Query OK, 0 rows affected (0.614 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> CREATE INDEX idx_orders_status
-> ON orders(order_status);
Query OK, 0 rows affected (0.399 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> CREATE TABLE order_items (
-> item_id BIGINT PRIMARY KEY,
-> order_id BIGINT,
-> product_id BIGINT,
-> quantity INT,
-> pay_amount DECIMAL(12,2)
-> );
Query OK, 0 rows affected (0.074 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> CREATE INDEX idx_items_order
-> ON order_items(order_id);
Query OK, 0 rows affected (0.390 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> CREATE INDEX idx_items_product
-> ON order_items(product_id);
Query OK, 0 rows affected (0.500 sec)
obclient(root@sys)[ob_test]> DELIMITER $$
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> CREATE PROCEDURE sp_insert_customers(IN cnt INT)
-> BEGIN
-> DECLARE i INT DEFAULT 1;
-> DECLARE v_region VARCHAR(32);
-> DECLARE v_level INT;
->
-> WHILE i <= cnt DO
-> SET v_region = CASE MOD(i, 4)
-> WHEN 0 THEN 'EAST'
-> WHEN 1 THEN 'WEST'
-> WHEN 2 THEN 'SOUTH'
-> ELSE 'NORTH'
-> END;
->
-> SET v_level = MOD(i, 5) + 1;
->
-> INSERT INTO customers(customer_id, region, level, created_at)
-> VALUES (
-> i,
-> v_region,
-> v_level,
-> DATE_SUB(CURDATE(), INTERVAL MOD(i, 365) DAY)
-> );
->
-> SET i = i + 1;
-> END WHILE;
-> END$$
Query OK, 0 rows affected (0.090 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> DELIMITER ;
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> CALL sp_insert_customers(10000);
Query OK, 1 row affected (14.091 sec)
obclient(root@sys)[ob_test]> DELIMITER $$
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> CREATE PROCEDURE sp_insert_products(IN cnt INT)
-> BEGIN
-> DECLARE i INT DEFAULT 1;
-> DECLARE v_category VARCHAR(32);
-> DECLARE v_status CHAR(1);
->
-> WHILE i <= cnt DO
-> SET v_category = CASE MOD(i, 5)
-> WHEN 0 THEN 'ELECTRONICS'
-> WHEN 1 THEN 'BOOK'
-> WHEN 2 THEN 'CLOTHING'
-> WHEN 3 THEN 'FOOD'
-> ELSE 'OTHER'
-> END;
->
-> SET v_status = IF(MOD(i, 10) < 8, 'A', 'I'); -- 80% 在售
->
-> INSERT INTO products(product_id, category, price, status)
-> VALUES (
-> i,
-> v_category,
-> ROUND(50 + MOD(i, 5000) * 0.1, 2),
-> v_status
-> );
->
-> SET i = i + 1;
-> END WHILE;
-> END$$
Query OK, 0 rows affected (0.036 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> DELIMITER ;
obclient(root@sys)[ob_test]> CALL sp_insert_products(5000);
Query OK, 1 row affected (8.025 sec)
obclient(root@sys)[ob_test]> DELIMITER $$
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> CREATE PROCEDURE sp_insert_orders(IN cnt INT)
-> BEGIN
-> DECLARE i INT DEFAULT 1;
-> DECLARE v_customer BIGINT;
-> DECLARE v_status CHAR(1);
->
-> WHILE i <= cnt DO
-> SET v_customer = MOD(i, 10000) + 1;
-> SET v_status = IF(MOD(i, 10) < 7, 'S', 'C'); -- 70% 成功
->
-> INSERT INTO orders(order_id, customer_id, order_date, order_status, total_amount)
-> VALUES (
-> i,
-> v_customer,
-> DATE_SUB('2024-12-31', INTERVAL MOD(i, 365) DAY),
-> v_status,
-> 0
-> );
->
-> SET i = i + 1;
-> END WHILE;
-> END$$
Query OK, 0 rows affected (0.045 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> DELIMITER ;
obclient(root@sys)[ob_test]> CALL sp_insert_orders(50000);
Query OK, 1 row affected (1 min 2.733 sec)
obclient(root@sys)[ob_test]> DELIMITER $$
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> CREATE PROCEDURE sp_insert_order_items(IN cnt INT)
-> BEGIN
-> DECLARE i INT DEFAULT 1;
-> DECLARE v_order BIGINT;
-> DECLARE v_product BIGINT;
-> DECLARE v_qty INT;
-> DECLARE v_pay DECIMAL(12,2);
->
-> WHILE i <= cnt DO
-> SET v_order = MOD(i, 50000) + 1;
-> SET v_product = MOD(i, 5000) + 1;
-> SET v_qty = MOD(i, 5) + 1;
-> SET v_pay = v_qty * (50 + MOD(i, 200) * 1.5);
->
-> INSERT INTO order_items(item_id, order_id, product_id, quantity, pay_amount)
-> VALUES (
-> i,
-> v_order,
-> v_product,
-> v_qty,
-> v_pay
-> );
->
-> -- 回写订单金额（制造真实聚合场景）
-> UPDATE orders
-> SET total_amount = total_amount + v_pay
-> WHERE order_id = v_order;
->
-> SET i = i + 1;
-> END WHILE;
-> END$$
Query OK, 0 rows affected (0.050 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> DELIMITER ;
obclient(root@sys)[ob_test]> CALL sp_insert_order_items(100000);
Query OK, 1 row affected (4 min 41.857 sec)
obclient(root@sys)[ob_test]> ANALYZE TABLE customers;
Query OK, 0 rows affected (0.195 sec)
obclient(root@sys)[ob_test]> ANALYZE TABLE products;
Query OK, 0 rows affected (0.078 sec)
obclient(root@sys)[ob_test]> ANALYZE TABLE orders;
Query OK, 0 rows affected (0.237 sec)
obclient(root@sys)[ob_test]> ANALYZE TABLE order_items;
Query OK, 0 rows affected (0.147 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> SELECT
-> (SELECT COUNT(*) FROM customers) AS customers,
-> (SELECT COUNT(*) FROM products) AS products,
-> (SELECT COUNT(*) FROM orders) AS orders,
-> (SELECT COUNT(*) FROM order_items) AS items;
+-----------+----------+--------+--------+
| customers | products | orders | items |
+-----------+----------+--------+--------+
| 10000 | 5000 | 50000 | 100000 |
+-----------+----------+--------+--------+
1 row inset (0.046 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> SELECT
-> (SELECT COUNT(*) FROM customers) AS customers,
-> (SELECT COUNT(*) FROM products) AS products,
-> (SELECT COUNT(*) FROM orders) AS orders,
-> (SELECT COUNT(*) FROM order_items) AS items;
+-----------+----------+--------+--------+
| customers | products | orders | items |
+-----------+----------+--------+--------+
| 10000 | 5000 | 50000 | 100000 |
+-----------+----------+--------+--------+
1 row inset (0.046 sec)
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]>
obclient(root@sys)[ob_test]> explain SELECT
-> o.order_id,
-> COUNT(DISTINCT oi.product_id) AS product_cnt,
-> SUM(oi.pay_amount) AS total_pay
-> FROM orders o
-> JOIN order_items oi
-> ON o.order_id = oi.order_id
-> WHERE o.order_status = 'S'
-> GROUP BY o.order_id
-> HAVING COUNT(DISTINCT oi.product_id) >= 3
-> AND SUM(oi.pay_amount) > 10000;
+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| Query Plan |
+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| ======================================================================== |
| |ID|OPERATOR |NAME |EST.ROWS|EST.TIME(us)| |
| ------------------------------------------------------------------------ |
| |0 |HASH GROUP BY | |392 |65992 | |
| |1 |└─SUBPLAN SCAN |VIEW1 |69255 |49429 | |
| |2 | └─HASH GROUP BY | |69255 |49245 | |
| |3 | └─HASH JOIN | |69262 |24937 | |
| |4 | ├─TABLE RANGE SCAN|o(idx_orders_status)|35000 |915 | |
| |5 | └─TABLE FULL SCAN |oi |100005 |6499 | |
| ======================================================================== |
| Outputs & filters: |
| ------------------------------------- |
| 0 - output([VIEW1.o.order_id], [T_FUN_COUNT(VIEW1.oi.product_id)], [cast(T_FUN_SUM(VIEW1.T_FUN_SUM(oi.pay_amount)), DECIMAL_INT(34, 2))]), filter([T_FUN_COUNT(VIEW1.oi.product_id) |
| >= 3], [cast(T_FUN_SUM(VIEW1.T_FUN_SUM(oi.pay_amount)), DECIMAL_INT(34, 2)) > cast(10000, DECIMAL_INT(34, 2))]), rowset=256 |
| group([VIEW1.o.order_id]), agg_func([T_FUN_SUM(VIEW1.T_FUN_SUM(oi.pay_amount))], [T_FUN_COUNT(VIEW1.oi.product_id)]) |
| 1 - output([VIEW1.T_FUN_SUM(oi.pay_amount)], [VIEW1.o.order_id], [VIEW1.oi.product_id]), filter(nil), rowset=256 |
| access([VIEW1.T_FUN_SUM(oi.pay_amount)], [VIEW1.o.order_id], [VIEW1.oi.product_id]) |
| 2 - output([T_FUN_SUM(oi.pay_amount)], [o.order_id], [oi.product_id]), filter(nil), rowset=256 |
| group([o.order_id], [oi.product_id]), agg_func([T_FUN_SUM(oi.pay_amount)]) |
| 3 - output([o.order_id], [oi.product_id], [oi.pay_amount]), filter(nil), rowset=256 |
| equal_conds([o.order_id = oi.order_id]), other_conds(nil) |
| 4 - output([o.order_id]), filter(nil), rowset=256 |
| access([o.order_id]), partitions(p0) |
| is_index_back=false, is_global_index=false, |
| range_key([o.order_status], [o.order_id]), range(S,MIN ; S,MAX), |
| range_cond([o.order_status = 'S']) |
| 5 - output([oi.order_id], [oi.product_id], [oi.pay_amount]), filter(nil), rowset=256 |
| access([oi.order_id], [oi.product_id], [oi.pay_amount]), partitions(p0) |
| is_index_back=false, is_global_index=false, |
| range_key([oi.item_id]), range(MIN ; MAX)always true |
+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
30 rows inset (0.017 sec)
obclient(root@sys)[ob_test]>
我们先分析上面的SQL如何进行处理， SELECT
-> o.order_id,
-> COUNT(DISTINCT oi.product_id) AS product_cnt,
-> SUM(oi.pay_amount) AS total_pay
-> FROM orders o
-> JOIN order_items oi
-> ON o.order_id = oi.order_id
-> WHERE o.order_status = 'S'
-> GROUP BY o.order_id
-> HAVING COUNT(DISTINCT oi.product_id) >= 3
-> AND SUM(oi.pay_amount) > 10000;
在SEEKDB 中首先我们一句话来分析执行计划中，先扫描订单，然后扫描明细表，通过hash join 将数据进行join后再进行聚合，having 过滤。这里我们添加推卸索引，优化后，再进行查询计划的分析。
obclient(root@sys)[ob_test]> explain SELECTo.order_id,
-> COUNT(DISTINCT oi.product_id) AS product_cnt,
-> SUM(oi.pay_amount) AS total_pay
-> FROM orders o
-> JOIN order_items oi
-> ON o.order_id = oi.order_id
-> WHERE o.order_status = 'S'
-> GROUP BY o.order_id
-> HAVING COUNT(DISTINCT oi.product_id) >= 3
-> AND SUM(oi.pay_amount) > 10000;
ERROR 1064 (42000): You have an error in your SQL syntax; check the manual that corresponds to your OceanBase version for the right syntax to use near ',
COUNT(DISTINCT oi.product_id) AS product_cnt,
SUM(oi.pay_amount) AS to' at line 1
obclient(root@sys)[ob_test]> explain SELECT o.order_id,
-> COUNT(DISTINCT oi.product_id) AS product_cnt,
-> SUM(oi.pay_amount) AS total_pay
-> FROM orders o
-> JOIN order_items oi
-> ON o.order_id = oi.order_id
-> WHERE o.order_status = 'S'
-> GROUP BY o.order_id
-> HAVING COUNT(DISTINCT oi.product_id) >= 3 AND SUM(oi.pay_amount) > 10000;
+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| Query Plan |
+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| =================================================================================== |
| |ID|OPERATOR |NAME |EST.ROWS|EST.TIME(us)| |
| ----------------------------------------------------------------------------------- |
| |0 |MERGE GROUP BY | |392 |39687 | |
| |1 |└─SUBPLAN SCAN |VIEW1 |69251 |32516 | |
| |2 | └─MERGE GROUP BY | |69251 |32332 | |
| |3 | └─MERGE JOIN | |69258 |22145 | |
| |4 | ├─TABLE FULL SCAN |oi(idx_items_order_product_pay)|100000 |6498 | |
| |5 | └─TABLE RANGE SCAN|o(idx_orders_status) |35000 |915 | |
| =================================================================================== |
| Outputs & filters: |
| ------------------------------------- |
| 0 - output([VIEW1.o.order_id], [T_FUN_COUNT(VIEW1.oi.product_id)], [cast(T_FUN_SUM(VIEW1.T_FUN_SUM(oi.pay_amount)), DECIMAL_INT(34, 2))]), filter([T_FUN_COUNT(VIEW1.oi.product_id) |
| >= 3], [cast(T_FUN_SUM(VIEW1.T_FUN_SUM(oi.pay_amount)), DECIMAL_INT(34, 2)) > cast(10000, DECIMAL_INT(34, 2))]), rowset=256 |
| group([VIEW1.o.order_id]), agg_func([T_FUN_SUM(VIEW1.T_FUN_SUM(oi.pay_amount))], [T_FUN_COUNT(VIEW1.oi.product_id)]) |
| 1 - output([VIEW1.T_FUN_SUM(oi.pay_amount)], [VIEW1.o.order_id], [VIEW1.oi.product_id]), filter(nil), rowset=256 |
| access([VIEW1.T_FUN_SUM(oi.pay_amount)], [VIEW1.o.order_id], [VIEW1.oi.product_id]) |
| 2 - output([T_FUN_SUM(oi.pay_amount)], [o.order_id], [oi.product_id]), filter(nil), rowset=256 |
| group([o.order_id], [oi.product_id]), agg_func([T_FUN_SUM(oi.pay_amount)]) |
| 3 - output([o.order_id], [oi.product_id], [oi.pay_amount]), filter(nil), rowset=256 |
| equal_conds([o.order_id = oi.order_id]), other_conds(nil) |
| merge_directions([ASC]) |
| 4 - output([oi.order_id], [oi.product_id], [oi.pay_amount]), filter(nil), rowset=256 |
| access([oi.order_id], [oi.product_id], [oi.pay_amount]), partitions(p0) |
| is_index_back=false, is_global_index=false, |
| range_key([oi.order_id], [oi.product_id], [oi.pay_amount], [oi.item_id]), range(MIN,MIN,MIN,MIN ; MAX,MAX,MAX,MAX)always true |
| 5 - output([o.order_id]), filter(nil), rowset=256 |
| access([o.order_id]), partitions(p0) |
| is_index_back=false, is_global_index=false, |
| range_key([o.order_status], [o.order_id]), range(S,MIN ; S,MAX), |
| range_cond([o.order_status = 'S']) |
+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
31 rows inset (0.013 sec)
优化前的执行计划
优化后的执行计划
从先后两个执行的执行计划看，整体执行计划有了较大的变化
1 Hash 变为 merge 那么这样有什么好处
首先这里的数据是有序的，在有对应的索引后，并不需要构建hash表，不需要更大的内存来进行数据的重构，在使用中边扫描边用相关的数据。
同时我们注意虽然还有table full scan 但是这里扫描的不是表扫描的是新加的索引，这里在顺序的扫描b+tree 速度比堆表要快的多。
2 优化后的执行顺序
2.1 oi(idx_items_order_product_pay) (order_id, product_id, pay_amount)
2.2 idx_orders_status (order_status='S', order_id)
2.3 o.order_id = oi.order_id
2.4 GROUP BY (order_id, product_id)
2.5 GROUP BY (order_id) HAVING ...
优化并后并不只是速度的快慢变化，而是对于内存的需求的大小的变化，有利于在有限的资源提供更好的SQL执行计划，这才是客户端，线下产品要的。
但是如果是MYSQL会怎么样，俩字 凉拌 ！！
1 线下客户端SQL的执行方式和服务器端完全不一样，没有足够的内存，CPU提供，没有专人去维护，我们需要的是皮实耐用，且通过简单优化就可以达到良好优化效果的数据库，而不是娇滴滴的，官宦小姐，没有专业的人去伺候她。
2 兼容MYSQL的语法，安装简单，同时优化器强悍，这就是线下客户端要的数据库产品
（别拿服务器端说事，要的是客户终端产品,简单批示耐用，才是王道）SeekDB需要仅仅是人气，和时间， 和纪老师的坚持，不信我们就走着瞧！！
和架构师沟通那种“一坨”的系统，推荐只能是OceanBase，Why ?
OceanBase Hybrid search 能力测试，平换MySQL的好选择
写了3750万字的我，在2000字的OB白皮书上了一课--记 《OceanBase 社区版在泛互场景的应用案例研究
OceanBase 6大学习法--OBCA视频学习总结第六章
OceanBase 6大学习法--OBCA视频学习总结第五章--索引与表设计
OceanBase 6大学习法--OBCA视频学习总结第五章--开发与库表设计
OceanBase 6大学习法--OBCA视频学习总结第四章 --数据库安装
OceanBase 6大学习法--OBCA视频学习总结第三章--数据库引擎
OceanBase 架构学习--OB上手视频学习总结第二章 （OBCA）
OceanBase 6大学习法--OB上手视频学习总结第一章
没有谁是垮掉的一代--记 第四届 OceanBase 数据库大赛
跟我学OceanBase4.0 --阅读白皮书 (OB分布式优化哪里了提高了速度)
跟我学OceanBase4.0 --阅读白皮书 （4.0优化的核心点是什么）
跟我学OceanBase4.0 --阅读白皮书 （0.5-4.0的架构与之前架构特点）
跟我学OceanBase4.0 --阅读白皮书 （旧的概念害死人呀，更新知识和理念）
OceanBase 学习记录-- 建立MySQL租户,像用MySQL一样使用OB
“合体吧兄弟们！”——从浪浪山小妖怪看OceanBase国产芯片优化《OceanBase “重如尘埃”之歌》
MongoDB “升级项目” 大型连续剧(3)-- 自动校对代码与注意事项
MongoDB “升级项目” 大型连续剧(2)-- 到底谁是"der"
MongoDB “升级项目” 大型连续剧(1)-- 可“生”可不升
MongoDB 大俗大雅，上来问分片真三俗 -- 4 分什么分
MongoDB 大俗大雅，高端知识讲“庸俗” --3 奇葩数据更新方法
MongoDB 大俗大雅，高端的知识讲“通俗” -- 2 嵌套和引用
MongoDB 大俗大雅，高端的知识讲“低俗” -- 1 什么叫多模
MongoDB 合作考试报销活动 贴附属，MongoDB基础知识速通
MongoDB 使用网上妙招，直接DOWN机---清理表碎片导致的灾祸 (送书活动结束)
MongoDB 2023年度纽约 MongoDB 年度大会话题 -- MongoDB 数据模式与建模
MongoDB 麻烦专业点，不懂可以问，别这么用行吗 ! --TTL
免费PolarDB云原生课程，听课“争”礼品，重塑云上知识，提高专业能力
非“厂商广告”的PolarDB课程：用户共创的新式学习范本--7位同学获奖PolarDB学习之星
“当复杂的SQL不再需要特别的优化”，邪修研究PolarDB for PG 列式索引加速复杂SQL运行
“PostgreSQL” 高性能主从强一致读写分离，我行，你没戏！
POLARDB 添加字段 “卡” 住---这锅Polar不背
PolarDB 版本差异分析--外人不知道的秘密（谁是绵羊，谁是怪兽）
PolarDB 答题拿-- 飞刀总的书、同款卫衣、T恤，来自杭州的Package（活动结束了）
PolarDB for MySQL 三大核心之一POLARFS 今天扒开它--- 嘛是火
PostgreSQL 新版本就一定好--由培训现象让我做的实验
说我PG Freezing Boom 讲的一般的那个同学，专帖给你，看看这次可满意
PostgreSQL 无服务 Neon and Aurora 新技术下的新经济模式 （翻译）
“PostgreSQL” 高性能主从强一致读写分离，我行，你没戏！
全世界都在“搞” PostgreSQL ，从Oracle 得到一个“馊主意”开始
PostgreSQL 加索引系统OOM 怨我了--- 不怨你怨谁
PostgreSQL “我怎么就连个数据库都不会建？” --- 你还真不会！
PostgreSQL 稳定性平台 PG中文社区大会--杭州来去匆匆
PostgreSQL 分组查询可以不进行全表扫描吗？速度提高上千倍？
POSTGRESQL --Austindatabaes 历年文章整理
PostgreSQL 查询语句开发写不好是必然，不是PG的锅
这个 PostgreSQL 让我有资本找老板要 鸡腿 鸭腿 !!
MySQL相关文章
一篇为MySQL用户，分析版本核心差异的文章--8.028-8.4的差异
那个MySQL大事务比你稳定，主从延迟低，为什么？ Look my eyes! 因为宋利兵宋老师