Split View: 데이터베이스 성능 튜닝 완전 가이드: 쿼리 최적화, 인덱스 전략, 커넥션 풀 관리까지

Bottleneck in the request flow:
Client -> CDN -> Load Balancer -> App Server -> [Database]
                                                  ^
                                           Most latency here

Performance improvement pyramid:

        /\
       /  \        Query optimization (greatest impact)
      /____\
     /      \      Index strategy
    /________\
   /          \    Schema design
  /____________\
 /              \  Hardware/config tuning
/________________\ Caching / Read replicas

-- PostgreSQL
EXPLAIN ANALYZE
SELECT u.name, o.total
FROM users u
JOIN orders o ON u.id = o.user_id
WHERE o.created_at > '2025-01-01'
  AND o.total > 100;

QUERY PLAN
---------------------------------------------------------------------------
Nested Loop  (cost=0.85..892.45 rows=23 width=48)
             (actual time=0.045..2.341 rows=19 loops=1)
  ->  Index Scan using idx_orders_created_at on orders o
        (cost=0.42..445.23 rows=23 width=20)
        (actual time=0.025..1.123 rows=19 loops=1)
        Filter: (total > 100)
        Rows Removed by Filter: 5
  ->  Index Scan using users_pkey on users u
        (cost=0.42..8.44 rows=1 width=36)
        (actual time=0.008..0.008 rows=1 loops=19)
Planning Time: 0.234 ms
Execution Time: 2.456 ms

-- 1. Seq Scan - reads the entire table
-- Appropriate for small tables or when most rows need reading
EXPLAIN ANALYZE SELECT * FROM small_table;

-- 2. Index Scan - finds row location via index, then accesses table
-- Suitable for high-selectivity (few rows returned) queries
EXPLAIN ANALYZE SELECT * FROM users WHERE email = 'user@example.com';

-- 3. Index Only Scan - returns results from index alone (no table access)
-- When using a Covering Index
EXPLAIN ANALYZE SELECT id, email FROM users WHERE email = 'user@example.com';

-- 4. Bitmap Index Scan - builds bitmap from index, then accesses table
-- Suitable for medium-selectivity queries
EXPLAIN ANALYZE SELECT * FROM orders WHERE status = 'pending' AND total > 100;

Scan type performance order (general):

Index Only Scan  >  Index Scan  >  Bitmap Scan  >  Seq Scan
(fastest)                                          (slowest)

-- PostgreSQL EXPLAIN options
EXPLAIN (ANALYZE, BUFFERS, FORMAT JSON)
SELECT * FROM orders WHERE user_id = 42;

-- BUFFERS: Include I/O information
-- FORMAT JSON: Output in JSON format
-- VERBOSE: Include additional details
-- COSTS: Include cost estimates (ON by default)
-- TIMING: Include timing info (ON by default)

-- MySQL EXPLAIN options
EXPLAIN ANALYZE
SELECT * FROM orders WHERE user_id = 42;

-- MySQL 8.0+ supports tree format
EXPLAIN FORMAT=TREE
SELECT * FROM orders WHERE user_id = 42;

-- Single column index
CREATE INDEX idx_users_email ON users(email);

-- Composite index (column order matters!)
CREATE INDEX idx_orders_user_date ON orders(user_id, created_at);

-- Composite index usage rules (Leftmost Prefix Rule)
-- idx_orders_user_date works for:
-- WHERE user_id = 1                          (O)
-- WHERE user_id = 1 AND created_at > '2025'  (O)
-- WHERE created_at > '2025'                  (X) missing leading column

-- Full-text search GIN index
CREATE INDEX idx_articles_search
ON articles USING GIN(to_tsvector('english', title || ' ' || body));

-- JSONB GIN index
CREATE INDEX idx_events_data ON events USING GIN(metadata);

-- JSONB specific path index
CREATE INDEX idx_events_type ON events USING GIN((metadata -> 'type'));

-- Usage examples
SELECT * FROM articles
WHERE to_tsvector('english', title || ' ' || body) @@ to_tsquery('postgresql & tuning');

SELECT * FROM events
WHERE metadata @> '{"type": "click", "page": "/home"}';

-- BRIN index - extremely efficient for time-series data
-- Over 100x smaller than B-Tree
CREATE INDEX idx_logs_created ON logs USING BRIN(created_at);

-- Effective when physical ordering strongly correlates
-- How to check:
SELECT correlation
FROM pg_stats
WHERE tablename = 'logs' AND attname = 'created_at';
-- correlation close to 1 means BRIN is suitable

-- Partial Index - indexes only rows matching a condition
CREATE INDEX idx_orders_pending
ON orders(created_at)
WHERE status = 'pending';
-- Only 5% of all rows indexed -> 95% size savings

-- Expression Index
CREATE INDEX idx_users_lower_email
ON users(LOWER(email));
-- Used by: WHERE LOWER(email) = 'user@example.com'

-- Covering Index
CREATE INDEX idx_orders_covering
ON orders(user_id, created_at) INCLUDE (total, status);
-- Enables Index Only Scan -> no table access needed

Index Design Checklist:

Must verify:
  [x] Indexes on columns frequently used in WHERE clauses
  [x] Indexes on JOIN condition columns
  [x] ORDER BY columns included in indexes
  [x] Column order optimization in composite indexes
  [x] High-selectivity columns first

Cautions:
  [x] Write vs read performance trade-off
  [x] Remove unused indexes
  [x] Monitor index size
  [x] Manage index bloat (PostgreSQL)

-- Enable pg_stat_statements (postgresql.conf)
-- shared_preload_libraries = 'pg_stat_statements'

-- Install extension
CREATE EXTENSION IF NOT EXISTS pg_stat_statements;

-- Top 10 slow queries (by total time)
SELECT
  substring(query, 1, 80) AS short_query,
  calls,
  round(total_exec_time::numeric, 2) AS total_ms,
  round(mean_exec_time::numeric, 2) AS mean_ms,
  round((100 * total_exec_time / sum(total_exec_time) OVER ())::numeric, 2) AS pct,
  rows
FROM pg_stat_statements
ORDER BY total_exec_time DESC
LIMIT 10;

-- Queries with highest average execution time
SELECT
  substring(query, 1, 100) AS short_query,
  calls,
  round(mean_exec_time::numeric, 2) AS mean_ms,
  rows / NULLIF(calls, 0) AS avg_rows
FROM pg_stat_statements
WHERE calls > 100
ORDER BY mean_exec_time DESC
LIMIT 10;

# my.cnf - Slow Query Log configuration
[mysqld]
slow_query_log = 1
slow_query_log_file = /var/log/mysql/slow-query.log
long_query_time = 1.0             # Log queries over 1 second
log_queries_not_using_indexes = 1 # Log queries not using indexes
min_examined_row_limit = 100      # Minimum 100 rows examined

# Analyze slow queries with mysqldumpslow
mysqldumpslow -s t -t 10 /var/log/mysql/slow-query.log

# pt-query-digest (Percona Toolkit)
pt-query-digest /var/log/mysql/slow-query.log \
  --limit 10 \
  --order-by query_time:sum

-- PostgreSQL: Check long-running queries in real time
SELECT
  pid,
  now() - pg_stat_activity.query_start AS duration,
  query,
  state
FROM pg_stat_activity
WHERE (now() - pg_stat_activity.query_start) > interval '5 seconds'
  AND state != 'idle'
ORDER BY duration DESC;

-- Check lock-waiting queries
SELECT
  blocked.pid AS blocked_pid,
  blocked.query AS blocked_query,
  blocking.pid AS blocking_pid,
  blocking.query AS blocking_query
FROM pg_stat_activity blocked
JOIN pg_locks bl ON bl.pid = blocked.pid
JOIN pg_locks kl ON kl.locktype = bl.locktype
  AND kl.database IS NOT DISTINCT FROM bl.database
  AND kl.relation IS NOT DISTINCT FROM bl.relation
JOIN pg_stat_activity blocking ON kl.pid = blocking.pid
WHERE NOT bl.granted AND kl.granted;

# N+1 problem example (Python/SQLAlchemy)

# BAD: N+1 queries triggered
users = session.query(User).all()  # 1 query
for user in users:
    print(user.orders)  # N additional queries (1 per user)

# Actual SQL executed:
# SELECT * FROM users;                    -- 1
# SELECT * FROM orders WHERE user_id = 1;  -- +1
# SELECT * FROM orders WHERE user_id = 2;  -- +1
# SELECT * FROM orders WHERE user_id = 3;  -- +1
# ... (N users)

# Solution 1: Eager Loading (JOIN)
users = session.query(User).options(
    joinedload(User.orders)
).all()
# SQL: SELECT * FROM users LEFT JOIN orders ON ...  -- 1 query

# Solution 2: Subquery Loading
users = session.query(User).options(
    subqueryload(User.orders)
).all()
# SQL: SELECT * FROM users;                         -- 1
# SQL: SELECT * FROM orders WHERE user_id IN (...);  -- 1

# Solution 3: selectinload (SQLAlchemy recommended)
users = session.query(User).options(
    selectinload(User.orders)
).all()

// Node.js (Prisma) - Solving N+1
// BAD: N+1
const users = await prisma.user.findMany();
for (const user of users) {
  const orders = await prisma.order.findMany({
    where: { userId: user.id }
  });
}

// GOOD: Solve with include
const users = await prisma.user.findMany({
  include: { orders: true }
});

// GOOD: Select only needed fields
const users = await prisma.user.findMany({
  select: {
    id: true,
    name: true,
    orders: {
      select: { id: true, total: true },
      where: { total: { gt: 100 } }
    }
  }
});

// Java (JPA/Hibernate)
// BAD: LAZY loading causes N+1
@Entity
public class User {
    @OneToMany(mappedBy = "user", fetch = FetchType.LAZY)
    private List<Order> orders;
}

// GOOD 1: JPQL JOIN FETCH
@Query("SELECT u FROM User u JOIN FETCH u.orders")
List<User> findAllWithOrders();

// GOOD 2: EntityGraph
@EntityGraph(attributePaths = {"orders"})
@Query("SELECT u FROM User u")
List<User> findAllWithOrders();

// GOOD 3: BatchSize (Hibernate)
@Entity
public class User {
    @OneToMany(mappedBy = "user")
    @BatchSize(size = 100)
    private List<Order> orders;
}

# Spring Boot - Hibernate N+1 detection
# application.yml
spring:
  jpa:
    properties:
      hibernate:
        generate_statistics: true
        session.events.log.LOG_QUERIES_SLOWER_THAN_MS: 25

logging:
  level:
    org.hibernate.SQL: DEBUG
    org.hibernate.stat: DEBUG

# Django - django-debug-toolbar or nplusone
# settings.py
INSTALLED_APPS = [
    'nplusone.ext.django',
]

NPLUSONE_RAISE = True  # Raise error on N+1 detection in dev

Without connection pool:
Request -> [Create new connection 30ms] -> [Query 5ms] -> [Close connection] = 35ms

With connection pool:
Request -> [Get existing connection from pool 0.1ms] -> [Query 5ms] -> [Return to pool] = 5.1ms

# application.yml - Optimal HikariCP settings
spring:
  datasource:
    hikari:
      # Pool size formula: connections = (core_count * 2) + spinning_disks
      # 4-core SSD server: (4 * 2) + 0 = 8
      maximum-pool-size: 10
      minimum-idle: 5

      # Connection timeout (max wait time from pool)
      connection-timeout: 30000   # 30 seconds

      # Idle connection removal time
      idle-timeout: 600000        # 10 minutes

      # Maximum connection lifetime
      max-lifetime: 1800000       # 30 minutes

      # Connection validation query
      connection-test-query: "SELECT 1"

      # Connection leak detection
      leak-detection-threshold: 60000  # 60 seconds

      # Pool name (for monitoring)
      pool-name: "MyApp-HikariPool"

# pgbouncer.ini
[databases]
myapp = host=localhost port=5432 dbname=myapp

[pgbouncer]
# Pooling mode
# session: Maintain session (default)
# transaction: Per-transaction (recommended)
# statement: Per-statement (limited)
pool_mode = transaction

# Pool size
default_pool_size = 20
min_pool_size = 5
reserve_pool_size = 5

# Connection limits
max_client_conn = 200
max_db_connections = 50

# Timeouts
server_idle_timeout = 600
client_idle_timeout = 300
query_timeout = 30
client_login_timeout = 60

# Authentication
auth_type = md5
auth_file = /etc/pgbouncer/userlist.txt

# Logging
log_connections = 1
log_disconnections = 1
log_pooler_errors = 1
stats_period = 60

# Admin interface
listen_addr = 0.0.0.0
listen_port = 6432
admin_users = pgbouncer

-- PgBouncer status check
SHOW pools;
SHOW stats;
SHOW servers;
SHOW clients;

-- PostgreSQL connection count
SELECT count(*) AS total_connections,
       count(*) FILTER (WHERE state = 'active') AS active,
       count(*) FILTER (WHERE state = 'idle') AS idle,
       count(*) FILTER (WHERE state = 'idle in transaction') AS idle_in_txn
FROM pg_stat_activity
WHERE backend_type = 'client backend';

-- Check maximum connections
SHOW max_connections;

-- Anti-pattern 1: SELECT *
-- BAD
SELECT * FROM users WHERE id = 1;
-- GOOD
SELECT id, name, email FROM users WHERE id = 1;

-- Anti-pattern 2: Function usage invalidates index
-- BAD (cannot use index)
SELECT * FROM users WHERE YEAR(created_at) = 2025;
-- GOOD (can use index)
SELECT * FROM users
WHERE created_at BETWEEN '2025-01-01' AND '2025-12-31';

-- Anti-pattern 3: LIKE with leading wildcard
-- BAD (cannot use index)
SELECT * FROM products WHERE name LIKE '%phone%';
-- GOOD (use full-text search index)
SELECT * FROM products
WHERE to_tsvector('english', name) @@ to_tsquery('phone');

-- Anti-pattern 4: UNION ALL instead of OR
-- BAD (inefficient index use)
SELECT * FROM orders WHERE user_id = 1 OR status = 'pending';
-- GOOD
SELECT * FROM orders WHERE user_id = 1
UNION ALL
SELECT * FROM orders WHERE status = 'pending' AND user_id != 1;

-- Anti-pattern 5: Large IN clause
-- BAD
SELECT * FROM products WHERE id IN (1, 2, 3, ..., 10000);
-- GOOD (use temp table or ANY)
SELECT * FROM products WHERE id = ANY(ARRAY[1, 2, 3, ...]);
-- Or use temp table JOIN

-- Anti-pattern 6: Unnecessary DISTINCT
-- BAD
SELECT DISTINCT u.name FROM users u JOIN orders o ON u.id = o.user_id;
-- GOOD (use EXISTS)
SELECT u.name FROM users u WHERE EXISTS (
  SELECT 1 FROM orders o WHERE o.user_id = u.id
);

-- Anti-pattern 7: COUNT(*) vs EXISTS
-- BAD (counts all rows)
SELECT CASE WHEN COUNT(*) > 0 THEN true ELSE false END
FROM orders WHERE user_id = 1;
-- GOOD (checks only first row)
SELECT EXISTS(SELECT 1 FROM orders WHERE user_id = 1);

-- Anti-pattern 8: Implicit type conversion
-- BAD (string column compared with number -> index invalidated)
SELECT * FROM users WHERE phone = 01012345678;
-- GOOD
SELECT * FROM users WHERE phone = '01012345678';

-- Anti-pattern 9: OFFSET pagination
-- BAD (large OFFSET is slow)
SELECT * FROM orders ORDER BY id LIMIT 20 OFFSET 100000;
-- GOOD (keyset pagination)
SELECT * FROM orders WHERE id > 100000 ORDER BY id LIMIT 20;

-- Anti-pattern 10: Excessive subqueries
-- BAD
SELECT * FROM users WHERE id IN (
  SELECT user_id FROM orders WHERE total > (
    SELECT AVG(total) FROM orders
  )
);
-- GOOD (CTE or JOIN)
WITH avg_total AS (SELECT AVG(total) AS avg_val FROM orders)
SELECT DISTINCT u.*
FROM users u
JOIN orders o ON u.id = o.user_id
CROSS JOIN avg_total
WHERE o.total > avg_total.avg_val;

-- JOIN order optimization (smaller table first)
-- PostgreSQL optimizes automatically, but when hints are needed:

-- Check table sizes
SELECT relname, reltuples::bigint AS row_count
FROM pg_class
WHERE relname IN ('users', 'orders', 'products')
ORDER BY reltuples DESC;

-- Use indexed columns in JOIN
-- GOOD
SELECT u.name, o.total
FROM orders o
JOIN users u ON u.id = o.user_id  -- users.id is PK (indexed)
WHERE o.status = 'completed';

-- Lateral Join usage (top N pattern)
SELECT u.name, recent_orders.*
FROM users u
CROSS JOIN LATERAL (
  SELECT id, total, created_at
  FROM orders
  WHERE user_id = u.id
  ORDER BY created_at DESC
  LIMIT 3
) recent_orders;

-- Range partitioning (ideal for time-series data)
CREATE TABLE orders (
  id bigserial,
  user_id bigint NOT NULL,
  total decimal(10,2),
  status varchar(20),
  created_at timestamp NOT NULL
) PARTITION BY RANGE (created_at);

-- Create monthly partitions
CREATE TABLE orders_2025_01
  PARTITION OF orders
  FOR VALUES FROM ('2025-01-01') TO ('2025-02-01');

CREATE TABLE orders_2025_02
  PARTITION OF orders
  FOR VALUES FROM ('2025-02-01') TO ('2025-03-01');

-- Automatic partition creation (using pg_partman)
-- CREATE EXTENSION pg_partman;
SELECT partman.create_parent(
  'public.orders',
  'created_at',
  'native',
  'monthly'
);

-- List partitioning (split by category)
CREATE TABLE products (
  id bigserial,
  name varchar(255),
  category varchar(50),
  price decimal(10,2)
) PARTITION BY LIST (category);

CREATE TABLE products_electronics
  PARTITION OF products
  FOR VALUES IN ('electronics', 'computers', 'phones');

CREATE TABLE products_clothing
  PARTITION OF products
  FOR VALUES IN ('clothing', 'shoes', 'accessories');

-- Hash partitioning (even distribution)
CREATE TABLE sessions (
  id uuid,
  user_id bigint,
  data jsonb,
  expires_at timestamp
) PARTITION BY HASH (user_id);

CREATE TABLE sessions_p0
  PARTITION OF sessions
  FOR VALUES WITH (MODULUS 4, REMAINDER 0);

CREATE TABLE sessions_p1
  PARTITION OF sessions
  FOR VALUES WITH (MODULUS 4, REMAINDER 1);

CREATE TABLE sessions_p2
  PARTITION OF sessions
  FOR VALUES WITH (MODULUS 4, REMAINDER 2);

CREATE TABLE sessions_p3
  PARTITION OF sessions
  FOR VALUES WITH (MODULUS 4, REMAINDER 3);

-- MySQL Range partitioning
CREATE TABLE orders (
  id BIGINT AUTO_INCREMENT,
  user_id BIGINT NOT NULL,
  total DECIMAL(10,2),
  created_at DATETIME NOT NULL,
  PRIMARY KEY (id, created_at)
) PARTITION BY RANGE (YEAR(created_at)) (
  PARTITION p2023 VALUES LESS THAN (2024),
  PARTITION p2024 VALUES LESS THAN (2025),
  PARTITION p2025 VALUES LESS THAN (2026),
  PARTITION p_future VALUES LESS THAN MAXVALUE
);

Write/Read split architecture:

       Write requests         Read requests
          |                       |
     [Primary]     ->      [Replica 1]
          |                 [Replica 2]
          |                 [Replica 3]
     Async replication

# Spring Boot - Read/Write split
spring:
  datasource:
    primary:
      url: jdbc:postgresql://primary:5432/myapp
      username: app_write
    replica:
      url: jdbc:postgresql://replica:5432/myapp
      username: app_read

# Django - Read/Write router
# settings.py
DATABASES = {
    'default': {
        'ENGINE': 'django.db.backends.postgresql',
        'HOST': 'primary.db.example.com',
        'NAME': 'myapp',
    },
    'replica': {
        'ENGINE': 'django.db.backends.postgresql',
        'HOST': 'replica.db.example.com',
        'NAME': 'myapp',
    }
}

DATABASE_ROUTERS = ['myapp.routers.ReadReplicaRouter']

# routers.py
class ReadReplicaRouter:
    def db_for_read(self, model, **hints):
        return 'replica'

    def db_for_write(self, model, **hints):
        return 'default'

# Redis caching pattern
import redis
import json
from functools import wraps

r = redis.Redis(host='localhost', port=6379, db=0)

# Cache-Aside (Lazy Loading) pattern
def cache_aside(key_prefix, ttl=3600):
    def decorator(func):
        @wraps(func)
        def wrapper(*args, **kwargs):
            cache_key = f"{key_prefix}:{args}:{kwargs}"

            # 1. Look up in cache
            cached = r.get(cache_key)
            if cached:
                return json.loads(cached)

            # 2. Query from DB
            result = func(*args, **kwargs)

            # 3. Store in cache
            r.setex(cache_key, ttl, json.dumps(result))

            return result
        return wrapper
    return decorator

@cache_aside("user", ttl=1800)
def get_user_profile(user_id):
    return db.query("SELECT * FROM users WHERE id = %s", [user_id])

# Write-Through pattern
def update_user(user_id, data):
    # 1. Update DB
    db.execute("UPDATE users SET name=%s WHERE id=%s", [data['name'], user_id])

    # 2. Update cache (or delete)
    cache_key = f"user:{user_id}"
    r.delete(cache_key)  # Cache Invalidation
    # or
    r.setex(cache_key, 1800, json.dumps(data))  # Cache Update

Caching strategy comparison:

Cache-Aside:
  Pros: Most common, flexible
  Cons: First request is a cache miss

Write-Through:
  Pros: Cache is always up-to-date
  Cons: Increased write latency

Write-Behind (Write-Back):
  Pros: Best write performance
  Cons: Risk of data loss

Read-Through:
  Pros: Cache logic is separated
  Cons: Implementation complexity

-- Manual VACUUM
VACUUM (VERBOSE) orders;

-- VACUUM FULL (table rewrite - causes lock!)
VACUUM FULL orders;

-- Update statistics
ANALYZE orders;

-- VACUUM + ANALYZE simultaneously
VACUUM ANALYZE orders;

# postgresql.conf - autovacuum settings
autovacuum = on
autovacuum_max_workers = 3
autovacuum_naptime = 60

# Run autovacuum when 20% of table is changed
autovacuum_vacuum_scale_factor = 0.2
autovacuum_vacuum_threshold = 50

# Run autoanalyze when 10% of table is changed
autovacuum_analyze_scale_factor = 0.1
autovacuum_analyze_threshold = 50

# autovacuum speed control (I/O load)
autovacuum_vacuum_cost_delay = 2ms
autovacuum_vacuum_cost_limit = 200

-- HOT Update conditions:
-- 1. Updated column has no index
-- 2. New row can be stored in the same page

-- Check HOT ratio
SELECT
  relname,
  n_tup_upd,
  n_tup_hot_upd,
  round(100.0 * n_tup_hot_upd / NULLIF(n_tup_upd, 0), 2) AS hot_pct
FROM pg_stat_user_tables
WHERE n_tup_upd > 0
ORDER BY n_tup_upd DESC;

-- To increase HOT ratio:
-- 1. Minimize indexes on frequently updated columns
-- 2. Lower fillfactor to reserve space on the same page
ALTER TABLE orders SET (fillfactor = 80);

-- Per-table I/O statistics
SELECT
  relname,
  seq_scan,
  idx_scan,
  n_tup_ins,
  n_tup_upd,
  n_tup_del,
  n_live_tup,
  n_dead_tup
FROM pg_stat_user_tables
ORDER BY seq_scan DESC;

-- Index usage rates
SELECT
  indexrelname,
  idx_scan,
  idx_tup_read,
  idx_tup_fetch,
  pg_size_pretty(pg_relation_size(indexrelid)) AS index_size
FROM pg_stat_user_indexes
ORDER BY idx_scan;

-- Unused indexes (candidates for removal)
SELECT
  indexrelname,
  idx_scan,
  pg_size_pretty(pg_relation_size(indexrelid)) AS size
FROM pg_stat_user_indexes
WHERE idx_scan = 0
  AND indexrelname NOT LIKE '%_pkey'
ORDER BY pg_relation_size(indexrelid) DESC;

-- Cache hit ratio
SELECT
  sum(heap_blks_read) AS heap_read,
  sum(heap_blks_hit) AS heap_hit,
  round(100.0 * sum(heap_blks_hit) / NULLIF(sum(heap_blks_hit) + sum(heap_blks_read), 0), 2) AS hit_pct
FROM pg_statio_user_tables;
-- 99% or above is ideal

# my.cnf - InnoDB optimization
[mysqld]
# Buffer Pool size (70-80% of total memory)
innodb_buffer_pool_size = 8G

# Buffer Pool instances (split when 8G or more)
innodb_buffer_pool_instances = 8

# Log file size
innodb_log_file_size = 1G
innodb_log_buffer_size = 64M

# I/O threads
innodb_read_io_threads = 4
innodb_write_io_threads = 4

# Concurrency
innodb_thread_concurrency = 0  # Automatic

# Flush method
innodb_flush_method = O_DIRECT
innodb_flush_log_at_trx_commit = 1  # Safe (2: performance priority)

# Temporary tables
tmp_table_size = 256M
max_heap_table_size = 256M

-- Buffer Pool status check
SHOW ENGINE INNODB STATUS\G

-- Buffer Pool hit ratio
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_ratio;
-- 99% or above is ideal

MySQL 8.0 Query Cache alternatives:

1. Application-level caching (Redis/Memcached)
2. ProxySQL query cache
3. MySQL Router caching
4. InnoDB Buffer Pool optimization

-- MySQL 8.0 Performance Schema usage
-- Slow query analysis
SELECT
  DIGEST_TEXT,
  COUNT_STAR AS exec_count,
  ROUND(SUM_TIMER_WAIT/1000000000, 2) AS total_ms,
  ROUND(AVG_TIMER_WAIT/1000000000, 2) AS avg_ms,
  SUM_ROWS_EXAMINED,
  SUM_ROWS_SENT
FROM performance_schema.events_statements_summary_by_digest
ORDER BY SUM_TIMER_WAIT DESC
LIMIT 10;

-- Index hints (when optimizer makes wrong choices)
SELECT * FROM orders USE INDEX (idx_orders_user_date)
WHERE user_id = 1 AND created_at > '2025-01-01';

SELECT * FROM orders FORCE INDEX (idx_orders_status)
WHERE status = 'pending';

-- MySQL 8.0 optimizer hints
SELECT /*+ NO_INDEX(orders idx_orders_status) */
  * FROM orders WHERE user_id = 1;

SELECT /*+ JOIN_ORDER(users, orders) */
  u.name, o.total
FROM users u JOIN orders o ON u.id = o.user_id;