Split View: MySQL 쿼리 최적화 실전 가이드: EXPLAIN 분석부터 인덱스 설계·슬로우 쿼리 튜닝까지

EXPLAIN SELECT u.name, o.total_amount
FROM users u
JOIN orders o ON u.id = o.user_id
WHERE u.status = 'active'
  AND o.created_at >= '2026-01-01';

-- const: Single row lookup by primary key
EXPLAIN SELECT * FROM users WHERE id = 1;
-- type: const, rows: 1

-- ref: Non-unique index lookup
EXPLAIN SELECT * FROM orders WHERE user_id = 100;
-- type: ref (when user_id has an index)

-- ALL: Full scan - must be optimized
EXPLAIN SELECT * FROM orders WHERE YEAR(created_at) = 2026;
-- type: ALL (index unusable due to function application)

EXPLAIN ANALYZE
SELECT u.name, COUNT(o.id) as order_count
FROM users u
JOIN orders o ON u.id = o.user_id
WHERE u.created_at >= '2026-01-01'
GROUP BY u.id
HAVING order_count > 5
ORDER BY order_count DESC
LIMIT 10;

-- Manually refresh statistics
ANALYZE TABLE users;
ANALYZE TABLE orders;

-- Creating a B-Tree index
CREATE INDEX idx_orders_user_created ON orders (user_id, created_at);

-- Queries where this index is effective
SELECT * FROM orders WHERE user_id = 100 AND created_at >= '2026-01-01';
SELECT * FROM orders WHERE user_id = 100 ORDER BY created_at DESC;
SELECT * FROM orders WHERE user_id IN (100, 200, 300);

-- Using Hash index on a MEMORY engine table
CREATE TABLE session_cache (
    session_id VARCHAR(64) NOT NULL,
    user_id INT NOT NULL,
    data JSON,
    INDEX USING HASH (session_id)
) ENGINE = MEMORY;

-- Check InnoDB Adaptive Hash Index status
SHOW ENGINE INNODB STATUS\G
-- Check hit rate in the Adaptive Hash Index section

-- Creating a Full-text index
ALTER TABLE articles ADD FULLTEXT INDEX ft_idx_content (title, body);

-- Natural language mode search
SELECT * FROM articles
WHERE MATCH(title, body) AGAINST('MySQL performance optimization' IN NATURAL LANGUAGE MODE);

-- Boolean mode search (AND, OR, NOT, etc.)
SELECT * FROM articles
WHERE MATCH(title, body) AGAINST('+MySQL +performance -PostgreSQL' IN BOOLEAN MODE);

-- Creating a Spatial index
CREATE TABLE locations (
    id INT AUTO_INCREMENT PRIMARY KEY,
    name VARCHAR(100),
    position POINT NOT NULL SRID 4326,
    SPATIAL INDEX (position)
);

-- Search within radius
SELECT name, ST_Distance_Sphere(position, ST_SRID(POINT(127.0276, 37.4979), 4326)) AS distance
FROM locations
WHERE ST_Distance_Sphere(position, ST_SRID(POINT(127.0276, 37.4979), 4326)) < 1000;

-- Composite index: (a, b, c)
CREATE INDEX idx_abc ON orders (status, user_id, created_at);

-- Queries that CAN use the index
SELECT * FROM orders WHERE status = 'completed';                          -- (a) used
SELECT * FROM orders WHERE status = 'completed' AND user_id = 100;        -- (a, b) used
SELECT * FROM orders WHERE status = 'completed' AND user_id = 100
  AND created_at >= '2026-01-01';                                         -- (a, b, c) all used

-- Queries that CANNOT use the index
SELECT * FROM orders WHERE user_id = 100;                                 -- only (b) - index not used
SELECT * FROM orders WHERE user_id = 100 AND created_at >= '2026-01-01';  -- (b, c) - index not used
SELECT * FROM orders WHERE created_at >= '2026-01-01';                    -- only (c) - index not used

-- Covering index design
CREATE INDEX idx_covering ON orders (user_id, status, total_amount);

-- This query can return results from index only (Using index)
EXPLAIN SELECT user_id, status, total_amount
FROM orders
WHERE user_id = 100 AND status = 'completed';

-- This query does NOT use covering index (SELECT * requires table data access)
EXPLAIN SELECT * FROM orders
WHERE user_id = 100 AND status = 'completed';

-- Bad design: Index on low-cardinality column only
CREATE INDEX idx_bad ON orders (status);  -- status has only 5-10 distinct values

-- Good design: Composite index matching query patterns
CREATE INDEX idx_good ON orders (user_id, status, created_at);

-- Design considering sorting
-- When WHERE user_id = ? ORDER BY created_at DESC is frequent
CREATE INDEX idx_sort ON orders (user_id, created_at DESC);

-- Bad: Correlated subquery (subquery executed for each row)
SELECT u.name, u.email,
  (SELECT COUNT(*) FROM orders o WHERE o.user_id = u.id) AS order_count
FROM users u
WHERE u.status = 'active';

-- Good: LEFT JOIN + GROUP BY
SELECT u.name, u.email, COUNT(o.id) AS order_count
FROM users u
LEFT JOIN orders o ON u.id = o.user_id
WHERE u.status = 'active'
GROUP BY u.id, u.name, u.email;

-- Alternative: Using a Derived Table
SELECT u.name, u.email, COALESCE(oc.cnt, 0) AS order_count
FROM users u
LEFT JOIN (
    SELECT user_id, COUNT(*) AS cnt
    FROM orders
    GROUP BY user_id
) oc ON u.id = oc.user_id
WHERE u.status = 'active';

-- Bad: Index underutilization due to OR condition
SELECT * FROM products
WHERE category_id = 10 OR brand_id = 20;

-- Good: Split using UNION ALL (each condition can use its own index)
SELECT * FROM products WHERE category_id = 10
UNION ALL
SELECT * FROM products WHERE brand_id = 20 AND category_id != 10;

-- Bad: Function applied to indexed column
SELECT * FROM orders WHERE YEAR(created_at) = 2026;
SELECT * FROM users WHERE LOWER(email) = 'user@example.com';

-- Good: Rewrite as range condition
SELECT * FROM orders
WHERE created_at >= '2026-01-01' AND created_at < '2027-01-01';

-- MySQL 8.0+: Expression Index (Functional Index)
CREATE INDEX idx_email_lower ON users ((LOWER(email)));
SELECT * FROM users WHERE LOWER(email) = 'user@example.com';

-- Bad: phone_number is VARCHAR but compared with a number
SELECT * FROM users WHERE phone_number = 01012345678;
-- MySQL converts phone_number to numeric -> index not used

-- Good: Type matching
SELECT * FROM users WHERE phone_number = '01012345678';

-- USE INDEX: Suggest a specific index (optimizer may ignore)
SELECT * FROM orders USE INDEX (idx_user_created)
WHERE user_id = 100 AND created_at >= '2026-01-01';

-- FORCE INDEX: Force a specific index (must use index instead of full scan)
SELECT * FROM orders FORCE INDEX (idx_user_created)
WHERE user_id = 100 AND created_at >= '2026-01-01';

-- IGNORE INDEX: Exclude a specific index
SELECT * FROM orders IGNORE INDEX (idx_status)
WHERE status = 'completed' AND user_id = 100;

-- Fix JOIN order
SELECT /*+ JOIN_ORDER(u, o) */
  u.name, o.total_amount
FROM users u
JOIN orders o ON u.id = o.user_id
WHERE u.status = 'active';

-- Specify index for a specific table
SELECT /*+ INDEX(o idx_user_created) */
  o.*
FROM orders o
WHERE o.user_id = 100;

-- Force Hash Join
SELECT /*+ HASH_JOIN(u, o) */
  u.name, o.total_amount
FROM users u
JOIN orders o ON u.id = o.user_id;

-- Parallel query execution (MySQL 8.0.14+)
SELECT /*+ SET_VAR(innodb_parallel_read_threads=4) */
  COUNT(*) FROM large_table;

-- Check current settings
SHOW VARIABLES LIKE 'optimizer_switch';

-- Disable specific optimizations at session level
SET SESSION optimizer_switch = 'index_merge_intersection=off';
SET SESSION optimizer_switch = 'derived_merge=off';
SET SESSION optimizer_switch = 'batched_key_access=on';

-- MySQL 8.0.31+: Histogram-based optimization
ANALYZE TABLE orders UPDATE HISTOGRAM ON user_id, status WITH 100 BUCKETS;

-- Enable slow query log
SET GLOBAL slow_query_log = 'ON';
SET GLOBAL long_query_time = 1;  -- Record queries taking more than 1 second
SET GLOBAL log_queries_not_using_indexes = 'ON';  -- Also record queries not using indexes
SET GLOBAL slow_query_log_file = '/var/log/mysql/slow-query.log';

-- Permanent configuration in my.cnf
-- [mysqld]
-- slow_query_log = 1
-- long_query_time = 1
-- log_queries_not_using_indexes = 1
-- slow_query_log_file = /var/log/mysql/slow-query.log
-- min_examined_row_limit = 1000

# Top 10 slowest queries
mysqldumpslow -s t -t 10 /var/log/mysql/slow-query.log

# Top 10 most frequent slow queries
mysqldumpslow -s c -t 10 /var/log/mysql/slow-query.log

# Filter by specific pattern
mysqldumpslow -s t -t 10 -g "orders" /var/log/mysql/slow-query.log

# Analyze slow log
pt-query-digest /var/log/mysql/slow-query.log

# Analyze specific time range
pt-query-digest --since="2026-03-10 00:00:00" --until="2026-03-11 00:00:00" \
  /var/log/mysql/slow-query.log

# Save results to file
pt-query-digest /var/log/mysql/slow-query.log > /tmp/slow-query-report.txt

-- Check current buffer pool status
SHOW VARIABLES LIKE 'innodb_buffer_pool%';
SHOW STATUS LIKE 'Innodb_buffer_pool%';

-- Calculate buffer pool hit rate
SELECT
  (1 - (
    (SELECT VARIABLE_VALUE FROM performance_schema.global_status WHERE VARIABLE_NAME = 'Innodb_buffer_pool_reads') /
    (SELECT VARIABLE_VALUE FROM performance_schema.global_status WHERE VARIABLE_NAME = 'Innodb_buffer_pool_read_requests')
  )) * 100 AS buffer_pool_hit_rate;
-- 99% or higher is desirable

# my.cnf configuration example
[mysqld]
innodb_buffer_pool_size = 32G
innodb_buffer_pool_instances = 8    # Buffer pool size / instances = 4GB each
innodb_buffer_pool_chunk_size = 1G  # Online resizing unit
innodb_log_file_size = 4G
innodb_log_buffer_size = 64M
innodb_io_capacity = 2000
innodb_io_capacity_max = 4000

-- Dump buffer pool at shutdown, load at startup
SET GLOBAL innodb_buffer_pool_dump_at_shutdown = ON;
SET GLOBAL innodb_buffer_pool_load_at_startup = ON;

-- Manual dump/load
SET GLOBAL innodb_buffer_pool_dump_now = ON;
SET GLOBAL innodb_buffer_pool_load_now = ON;

-- Check load progress
SHOW STATUS LIKE 'Innodb_buffer_pool_load_status';

-- MySQL 8.4: EXPLAIN improvement example
EXPLAIN FORMAT=TREE
SELECT u.name, COUNT(o.id)
FROM users u
JOIN orders o ON u.id = o.user_id
GROUP BY u.id;

-- Using Invisible Index (test before dropping an index)
ALTER TABLE orders ALTER INDEX idx_status INVISIBLE;
-- After performance testing, if no issues, drop it
-- DROP INDEX idx_status ON orders;
-- If issues arise, reactivate
ALTER TABLE orders ALTER INDEX idx_status VISIBLE;

-- Problem: Numeric comparison on a VARCHAR column
-- account_no column is VARCHAR(20) but application passes a number
EXPLAIN SELECT * FROM accounts WHERE account_no = 123456;
-- type: ALL (full table scan!)

-- Cause: MySQL converts account_no column to numeric, making index unusable
-- Fix: Match the types
EXPLAIN SELECT * FROM accounts WHERE account_no = '123456';
-- type: ref (index used)

-- Problem: Index exists but optimizer chooses full scan
-- status column has an index but 'active' accounts for 90% of all rows
EXPLAIN SELECT * FROM users WHERE status = 'active';
-- type: ALL (optimizer determines full scan is more efficient)

-- Cause: Low cardinality makes the index inefficient
-- Fix 1: Create a composite index with other conditions
CREATE INDEX idx_status_created ON users (status, created_at);

-- Fix 2: When the index is only useful for specific values
-- MySQL does not directly support Partial Indexes, so consider changing query patterns

-- Problem: Index created as (created_at, user_id)
-- Frequent query: WHERE user_id = ? AND created_at >= ?
EXPLAIN SELECT * FROM orders
WHERE user_id = 100 AND created_at >= '2026-01-01';
-- Inefficient index usage (range scan on created_at first)

-- Fix: Place equality condition columns first
DROP INDEX idx_created_user ON orders;
CREATE INDEX idx_user_created ON orders (user_id, created_at);
-- type: range (efficient range scan)

-- Problem: Query performance degradation after deleting 5 million rows
-- Cause: Index pages have many empty spaces (fragmentation)

-- Diagnosis
SELECT
  TABLE_NAME,
  INDEX_LENGTH,
  DATA_LENGTH,
  DATA_FREE
FROM INFORMATION_SCHEMA.TABLES
WHERE TABLE_NAME = 'orders';

-- Fix: Rebuild indexes
ALTER TABLE orders ENGINE=InnoDB;  -- Rebuild table + indexes
-- Or
OPTIMIZE TABLE orders;

-- Find unused indexes
SELECT
  s.TABLE_SCHEMA,
  s.TABLE_NAME,
  s.INDEX_NAME,
  s.COLUMN_NAME
FROM INFORMATION_SCHEMA.STATISTICS s
LEFT JOIN performance_schema.table_io_waits_summary_by_index_usage p
  ON s.TABLE_SCHEMA = p.OBJECT_SCHEMA
  AND s.TABLE_NAME = p.OBJECT_NAME
  AND s.INDEX_NAME = p.INDEX_NAME
WHERE p.COUNT_STAR = 0
  AND s.TABLE_SCHEMA NOT IN ('mysql', 'information_schema', 'performance_schema', 'sys')
  AND s.INDEX_NAME != 'PRIMARY'
ORDER BY s.TABLE_SCHEMA, s.TABLE_NAME;

-- Find duplicate indexes (using sys schema)
SELECT * FROM sys.schema_redundant_indexes;

-- Check lock mode when adding an index
ALTER TABLE orders ADD INDEX idx_new (col1, col2), ALGORITHM=INPLACE, LOCK=NONE;

-- Use pt-online-schema-change (recommended for large tables)
-- Synchronizes the original table via triggers while making schema changes

pt-online-schema-change \
  --alter "ADD INDEX idx_new (col1, col2)" \
  --execute \
  D=mydb,t=orders

-- Create histogram
ANALYZE TABLE orders UPDATE HISTOGRAM ON status WITH 100 BUCKETS;
ANALYZE TABLE orders UPDATE HISTOGRAM ON total_amount WITH 254 BUCKETS;

-- Check histogram
SELECT
  SCHEMA_NAME,
  TABLE_NAME,
  COLUMN_NAME,
  JSON_EXTRACT(HISTOGRAM, '$.histogram-type') AS histogram_type,
  JSON_EXTRACT(HISTOGRAM, '$.number-of-buckets-specified') AS buckets
FROM INFORMATION_SCHEMA.COLUMN_STATISTICS;

-- Drop histogram
ANALYZE TABLE orders DROP HISTOGRAM ON status;