Redis Installation and Setup

Redis 5.0+ requires GCC 9.0 or higher for compilation.

# Download and extract
wget http://download.redis.io/releases/redis-3.2.12.tar.gz
tar xzf redis-3.2.12.tar.gz
mv redis-3.2.12 /data/redis

# Build from source
cd /data/redis
make

# Configure environment
vim /etc/profile
export PATH=/data/redis/src:$PATH
source /etc/profile

# Launch server
redis-server &

# Test connection
redis-cli
127.0.0.1:6379> set counter 10
OK
127.0.0.1:6379> get counter
10

Basic Configuration

mkdir /data/6379
cat > /data/6379/redis.conf <<'EOF'
daemonize yes
port 6379
logfile /data/6379/redis.log
dir /data/6379
dbfilename dump.rdb
EOF

redis-cli shutdown
redis-server /data/6379/redis.conf
netstat -lnp|grep 6379

Configuration Parameters

Parameter	Description
daemonize	Run in background (yes/no)
port	Listening port (default: 6379)
logfile	Log file location
dir	Data persistence directory
dbfilename	RDB snapshot filename

redis-cli
127.0.0.1:6379> set username john
OK
127.0.0.1:6379> get username
john

Systemd Service Configuration

[Unit]
Description=Redis
After=network.target

[Service]
ExecStart=/opt/redis/src/redis-server /etc/redis/redis.conf --daemonize no
ExecStop=/opt/redis/src/redis-cli -h 127.0.0.1 -p 6379 shutdown

[Install]
WantedBy=multi-user.target

Save as /lib/systemd/system/redis.service

Service Configuration File

# /etc/redis/redis.conf
daemonize yes
port 6379
logfile /var/log/redis/redis.log
dir /data/redis
dbfilename dump.rdb

mkdir -p /data/redis/ /var/log/redis/

Security Configuration

Redis enables protected mode by default, restricting access to localhost only.

Bind to Specific IPs

vim /data/6379/redis.conf
bind 10.0.0.51 127.0.0.1

Set Authentication Password

vim /data/6379/redis.conf
requirepass mysecretpass

Authentication Methods

# Method 1: Password in command
redis-cli -a mysecretpass
127.0.0.1:6379> set username john
OK
127.0.0.1:6379> exit

# Method 2: Interactive auth
redis-cli
127.0.0.1:6379> auth mysecretpass
OK
127.0.0.1:6379> set x y

# Remote authentication
redis-cli -a mysecretpass -h 10.0.0.51 -p 6379
10.0.0.51:6379> set data 2
OK

Runtime Configuration Management

CONFIG GET *
CONFIG GET requirepass
CONFIG GET r*
CONFIG SET requirepass newpassword

Data Persistence

Redis provides two persistence mechanisms: RDB and AOF.

RDB Persistence

Creates point-in-time snapshots at configured intervals.

Advantages: Fast backup, ideal for disaster recovery; master-replica sync depends on this mechanism.

Disadvantages: Potential data loss between snapshots.

vim /data/6379/redis.conf
dir /data/6379
dbfilename dump.rdb
save 900 1
save 300 10
save 60 10000

Configuration meanings:

• 1 change within 900 seconds (15 minutes)
• 10 changes within 300 seconds (5 minutes)
• 10000 changes within 60 seconds

AOF Persistence

Records all write operations as an append-only log file.

Advantages: Maximum data safety, similar to MySQL binary logging.

Disadvantages: Larger log files, potentially slower performance.

vim /data/6379/redis.conf
appendonly yes
appendfsync everysec

Sync strategies:

• always: Sync after every command
• everysec: Sync once per second (default)
• no: Let the OS handle syncing

Interview summary: RDB uses snapshots for faster backup; AOF logs operations sequentially for maximum data integrity.

Data Types

String     - String values
Hash       - Field-value pairs (dictionary)
List       - Ordered sequences
Set        - Unordered unique collections
Sorted Set - Score-ordered unique collections

Key Operations

KEYS pattern       # Pattern matching (KEYS *, KEYS a*, etc.)
TYPE key           # Value type
EXPIRE key seconds # Set TTL
PEXPIRE key ms     # Set TTL in milliseconds
TTL key            # Remaining time in seconds
PTTL key           # Remaining time in milliseconds
PERSIST key        # Remove TTL
DEL key            # Delete key
EXISTS key         # Check existence
RENAME old new     # Rename key

Examples:

127.0.0.1:6379> set username alice
127.0.0.1:6379> EXPIRE username 60
(integer) 1
127.0.0.1:6379> ttl username
(integer) 57
127.0.0.1:6379> set temp value EX 60
OK
127.0.0.1:6379> ttl temp
(integer) 60
127.0.0.1:6379> PERSIST temp
(integer) 1
127.0.0.1:6379> ttl temp
(integer) -1

String Operations

Use cases: Session storage, counters, caching.

# Basic operations
SET key value
MSET id 101 name bob age 25 gender m

# Counter operations
INCR counter           # Increment by 1
INCRBY counter 10000  # Increment by specified value
DECRBY counter 5000   # Decrement by specified value

# Extended operations
SETEX key 10 value          # Set with expiration (seconds)
SETNX key value             # Set if not exists
GETSET key newval           # Get old value and set new value
APPEND key suffix           # Append to existing value
STRLEN key                  # String length
SETRANGE key 5 newval       # Replace bytes starting at offset
GETRANGE key 0 10           # Substring extraction

Counter example:

127.0.0.1:6379> incr follower_count
(integer) 1
127.0.0.1:6379> get follower_count
1
127.0.0.1:6379> incrby follower_count 10000
(integer) 10006
127.0.0.1:6379> get follower_count
10006

Hash Operations

Use cases: User profiles, object caching, near-database performance.

# Storage
HMSET user:id:101 id 101 name alice age 25 gender f
HMSET user:id:102 id 102 name bob age 30 gender m

# Retrieval
HMGET user:id:101 id name age gender
HMGET user:id:102 id name age gender

Extended hash commands:

HSET hash field value      # Set single field
HSETNX hash field value    # Set if field doesn't exist
HGET hash field            # Get single field
HMGET hash f1 f2 f3        # Get multiple fields
HGETALL hash               # All fields and values
HLEN hash                  # Field count
HEXISTS hash field         # Field existence check
HKEYS hash                 # All field names
HVALS hash                 # All values
HINCRBY hash field delta    # Increment numeric value
HDEL hash field            # Delete field
DEL hash                   # Delete entire hash

Bulk data import example:

SELECT CONCAT('HMSET user_', id, ' id ', id, ' name ', name) 
FROM users INTO OUTFILE '/tmp/hash_import.txt'

List Operations

Use cases: Message queues, activity feeds, recent items.

# Add elements
LPUSH queue item1 item2     # Push to head
RPUSH queue item3 item4    # Push to tail
LPUSHX queue item          # Push if list exists
RPUSHX queue item          # Push to tail if list exists
LINSERT queue BEFORE pivot newval
LINSERT queue AFTER pivot newval

# Remove elements
LPOP queue                 # Pop from head
RPOP queue                 # Pop from tail
LREM queue count value     # Remove count occurrences
LTRIM queue 0 9            # Keep specified range

# Query elements
LRANGE queue 0 -1          # All elements
LINDEX queue position      # Element at index
LLEN queue                 # List length

WeChat timeline example:

127.0.0.1:6379> LPUSH timeline "weekend vibes!"
127.0.0.1:6379> LPUSH timeline "work hard"
127.0.0.1:6379> LPUSH timeline "coffee time"
127.0.0.1:6379> LPUSH timeline "Monday blues"
127.0.0.1:6379> LPUSH timeline "TGIF!"

127.0.0.1:6379> LRANGE timeline 0 0
1) "TGIF!"
127.0.0.1:6379> LRANGE timeline 0 2
1) "TGIF!"
2) "Monday blues"
3) "coffee time"
127.0.0.1:6379> LRANGE timeline -2 -1
1) "work hard"
2) "weekend vibes!"

Set Operations

Use cases: Social connections, tags, unique tracking.

# Add/remove members
SADD myset member1 member2 member3
SREM myset member1
SPOP myset                 # Remove and return random member

# Query operations
SMEMBERS myset            # All members
SISMEMBER myset member    # Existence check
SCARD myset               # Member count
SRANDMEMBER myset [count] # Random members without removal

# Set operations
SDIFF set1 set2           # Difference
SDIFFSTORE result set1 set2
SINTER set1 set2          # Intersection
SINTERSTORE result set1 set2
SUNION set1 set2          # Union
SUNIONSTORE result set1 set2

Social graph example:

127.0.0.1:6379> SADD following alice bob charlie david eve
(integer) 5
127.0.0.1:6379> SADD followers bob alice mike frank tony
(integer) 5

127.0.0.1:6379> SUNION following followers
1) "alice"
2) "bob"
3) "charlie"
4) "david"
5) "eve"
6) "mike"
7) "frank"
8) "tony"

127.0.0.1:6379> SINTER following followers
1) "alice"
2) "bob"

127.0.0.1:6379> SDIFF followers following
1) "tony"
2) "frank"
3) "mike"

Sorted Set Operations

Use cases: Leaderboards, ranking systems, priority queues.

# Add/update members
ZADD leaderboard score member
ZINCRBY leaderboard delta member

# Remove members
ZREM leaderboard member1 member2

# Query operations
ZRANGE leaderboard 0 -1 WITHSCORES           # Low to high
ZREVRANGE leaderboard 0 -1 WITHSCORES         # High to low
ZRANK leaderboard member                      # Position (low to high)
ZREVRANK leaderboard member                   # Position (high to low)
ZSCORE leaderboard member                      # Get score
ZCARD leaderboard                              # Member count
ZCOUNT leaderboard min max                     # Score range count
ZRANGEBYSCORE leaderboard 100 500              # Score range query
ZRANGEBYSCORE leaderboard -inf +inf LIMIT 0 10
ZREMRANGEBYSCORE leaderboard 0 100             # Remove by score
ZREMRANGEBYRANK leaderboard 0 2                # Remove by rank

Leaderboard example:

127.0.0.1:6379> ZADD rankings 0 player1 0 player2 0 player3 0 player4 0 player5 0 player6
(integer) 6
127.0.0.1:6379> ZINCRBY rankings 85000 player1
85000
127.0.0.1:6379> ZINCRBY rankings 92000 player2
92000
127.0.0.1:6379> ZINCRBY rankings 120000 player3
120000
127.0.0.1:6379> ZINCRBY rankings 1000 player4
1000

127.0.0.1:6379> ZREVRANGE rankings 0 2 WITHSCORES
1) "player3"
2) "120000"
3) "player2"
4) "92000"
5) "player1"
6) "85000"

Publish/Subscribe

PUBLISH channel message
SUBSCRIBE channel [channel ...]
UNSUBSCRIBE [channel ...]
PSUBSCRIBE pattern [pattern ...]
PUNSUBSCRIBE [pattern ...]
PUBSUB subcommand [argument ...]

Note: Subscribers receive only messages published after subscription. Offline messages are not buffered.

Basic example:

# Terminal 1 - Subscriber
SUBSCRIBE news

# Terminal 2 - Publisher
PUBLISH news "Breaking: Redis 7.0 released!"

Pattern subscription:

# Subscriber
PSUBSCRIBE news.*

# Publisher
PUBLISH news.tech "Tech updates"
PUBLISH news.sports "Sports news"

Transactions

Redis transactions use optimistic locking with WATCH/MULTI/EXEC.

MULTI
SET key1 value1
SET key2 value2
INCR counter
EXEC
# Or DISCARD to cancel

Execution behavior: Commands are queued and executed atomically. All succeed or all fail.

127.0.0.1:6379> MULTI
OK
127.0.0.1:6379> SET a 100
QUEUED
127.0.0.1:6379> SET b 200
QUEUED
127.0.0.1:6379> EXEC
1) OK
2) OK

Optimistic locking (ticket reservation):

# Initial ticket
SET ticket_count 1

# Terminal 1
WATCH ticket_count
MULTI
SET ticket_count 0

# Terminal 2
MULTI
SET ticket_count 0
EXEC

# Terminal 1
EXEC

Server Management Commands

INFO                   # Server statistics
CLIENT LIST            # Connected clients
CLIENT KILL ip:port    # Disconnect client
CONFIG GET *           # All configurations
CONFIG RESETSTAT       # Reset statistics
CONFIG SET param value # Change config at runtime
DBSIZE                 # Key count
FLUSHALL               # Clear all databases
FLUSHDB                # Clear current database
SELECT db_num          # Switch database
MONITOR                # Real-time command monitoring
SHUTDOWN               # Graceful shutdown

# Shutdown example
redis-cli -a password shutdown

Master-Replica Configuration

Replication Flow

1. Replica connects to master and sends SYNC
2. Master executes BGSAVE and sends RDB snapshot
3. Replica loads RDB snapshot
4. Master streams new operations to replica
5. Ongoing: master pushes all writes to replicas
6. Reconnection: replica sends PSYNC for incremental sync
7. Only missing data is transferred on reconnect

Consistency Safeguards

min-slaves-to-write 1
min-slaves-max-lag 3

Persistence Requirement

Critical: Master must have persistence enabled. Without it, master restart causes complete data loss across all replicas.

Setup Procedure

# Create directories
mkdir /data/638{0..2}

# Master configuration (6380)
cat > /data/6380/redis.conf <<EOF
port 6380
daemonize yes
pidfile /data/6380/redis.pid
loglevel notice
logfile /data/6380/redis.log
dbfilename dump.rdb
dir /data/6380
requirepass masterpass
masterauth masterpass
EOF

# Replica configuration (6381)
cat > /data/6381/redis.conf <<EOF
port 6381
daemonize yes
pidfile /data/6381/redis.pid
loglevel notice
logfile /data/6381/redis.log
dbfilename dump.rdb
dir /data/6381
requirepass replicapass
masterauth masterpass
EOF

# Replica configuration (6382)
cat > /data/6382/redis.conf <<EOF
port 6382
daemonize yes
pidfile /data/6382/redis.pid
loglevel notice
logfile /data/6382/redis.log
dbfilename dump.rdb
dir /data/6382
requirepass replicapass
masterauth masterpass
EOF

# Start instances
redis-server /data/6380/redis.conf
redis-server /data/6381/redis.conf
redis-server /data/6382/redis.conf

# Configure replication
redis-cli -p 6381 -a replicapass SLAVEOF 127.0.0.1 6380
redis-cli -p 6382 -a replicapass SLAVEOF 127.0.0.1 6380

# Verify status
redis-cli -p 6380 -a masterpass INFO replication
redis-cli -p 6381 -a replicapass INFO replication
redis-cli -p 6382 -a replicapass INFO replication

Redis Sentinel

Sentinel provides automatic failover with these capabilities:

• Continuous monitoring
• Automatic master election
• Automatic replica reassignment
• Transparent client redirection
• Automatic node failure handling

Sentinel Setup

mkdir /data/26380

cat > /data/26380/sentinel.conf <<EOF
port 26380
dir "/data/26380"
sentinel monitor mymaster 127.0.0.1 6380 1
sentinel down-after-milliseconds mymaster 5000
sentinel auth-pass mymaster masterpass
EOF

# Launch sentinel
redis-sentinel /data/26380/sentinel.conf &

Troubleshooting

1. Recreate 1-master-2-replica environment
2. Terminate sentinel processes
3. Clear sentinel directory
4. Reinitialize sentinel

Sentinel Commands

redis-cli -p 26380
PING                          # Returns PONG
SENTINEL masters              # List monitored masters
SENTINEL slaves <master>      # List replicas
SENTINEL get-master-addr-by-name <master>
SENTINEL reset <pattern>
SENTINEL failover <master>    # Manual failover

Redis Cluster

Architecture

High performance through distributed slots:

• 16384 slots distributed across master nodes
• Key slot determined by CRC16(key) % 16384
• Automatic client redirection to correct node

High availability through master-slave pairs:

• Automatic failover when master fails
• Each master has a replica for redundancy
• Typical deployment: 3 masters with 3 replicas

Slot distribution:

• Node 1: slots 0-5500
• Node 2: slots 5501-11000
• Node 3: slots 11001-16383

Installation

# Install Ruby dependencies
yum install ruby rubygems -y

# Configure Ruby gems
gem sources -l
gem sources -a http://mirrors.aliyun.com/rubygems/
gem sources --remove https://rubygems.org/
gem install redis -v 3.3.3

Node Preparation

mkdir /data/700{0..5}

# Generate configurations for ports 7000-7005
for port in 7000 7001 7002 7003 7004 7005; do
cat > /data/${port}/redis.conf <<EOF
port ${port}
daemonize yes
pidfile /data/${port}/redis.pid
loglevel notice
logfile /data/${port}/redis.log
dbfilename dump.rdb
dir /data/${port}
protected-mode no
cluster-enabled yes
cluster-config-file nodes.conf
cluster-node-timeout 5000
appendonly yes
EOF
done

# Start all nodes
for port in 7000 7001 7002 7003 7004 7005; do
  redis-server /data/${port}/redis.conf
done

# Verify processes
ps -ef | grep redis-server

Cluster Creation

redis-trib.rb create --replicas 1 \
  127.0.0.1:7000 127.0.0.1:7001 \
  127.0.0.1:7002 127.0.0.1:7003 \
  127.0.0.1:7004 127.0.0.1:7005

Cluster Management

Add new master:

# Create new directories and configs
mkdir /data/7006 /data/7007

# Start new nodes
redis-server /data/7006/redis.conf
redis-server /data/7007/redis.conf

# Add master to cluster
redis-trib.rb add-node 127.0.0.1:7006 127.0.0.1:7000

# Reshard slots to new master
redis-trib.rb reshard 127.0.0.1:7000

Add replica:

redis-trib.rb add-node --slave \
  --master-id <master-node-id> \
  127.0.0.1:7007 127.0.0.1:7000

Remove node:

# Reshard all slots away from target node first
redis-trib.rb reshard 127.0.0.1:7000

# Delete the node
redis-trib.rb del-node 127.0.0.1:7006 <node-id>

Memory management:

CONFIG SET maxmemory 102400000

Python Client Integration

Installation

yum install -y python36 python36-pip
pip3 install redis
pip3 install redis-py-cluster

Single Instance Connection

import redis

r = redis.StrictRedis(host='10.0.0.51', port=6379, db=0, password='mypassword')
r.set('username', 'alice')
print(r.get('username'))

Sentinel Connection

from redis.sentinel import Sentinel

sentinel = Sentinel([('localhost', 26380)], socket_timeout=0.1)

# Discover master and slaves
master_info = sentinel.discover_master('mymaster')
slave_info = sentinel.discover_slaves('mymaster')

# Create connection pools
master = sentinel.master_for('mymaster', socket_timeout=0.1, password='mypassword')
slave = sentinel.slave_for('mymaster', socket_timeout=0.1, password='mypassword')

# Read-write splitting
master.set('session:123', 'data_value')
result = slave.get('session:123')

Cluster Connection

from rediscluster import StrictRedisCluster

startup_nodes = [
    {"host": "127.0.0.1", "port": "7000"},
    {"host": "127.0.0.1", "port": "7001"},
    {"host": "127.0.0.1", "port": "7002"}
]

rc = StrictRedisCluster(startup_nodes=startup_nodes, decode_responses=True)
rc.set("cache_key", "cache_value")
print(rc.get("cache_key"))

Cache Design Patterns

Cache Penetration

Problem: Requests for non-existent keys bypass cache and hit database directly.

Soluitons:

• Bloom filter: Maintain a bitmap of possible keys; reject non-existent keys early
• Null caching: Store null values with short TTL for missing keys

Cache Avalanche

Problem: Mass keys expire simultaneously, causing database overload.

Solutions:

• Add random jitter to TTL values: actual_ttl = base_ttl + random(0, 300)
• Stagger expiration times across key groups

Cache Breakdown

Problem: A single hot key expires; concurrent requests all hit database.

Solutions:

• Use distributed locking: SETNX lock:key value EX 10
• Only the lock holder queries database; others wait and read from cache
• Release lock after cache population