Redis6.2.6 production environment stand-alone configuration

Detailed explanation of Redis production environment configuration

In actual production environments, in order to ensure the stability and high performance of Redis, we often need to make a series of optimizations to the default configuration. This blog post will take an optimized Redis configuration file as an example to explain in detail from multiple aspects such as basic network settings, persistence, replication, memory management, AOF persistence, performance tuning, and monitoring restrictions.

Here is a sample configuration file with detailed notes:

Configuration file example

#####################
# Redis configuration optimization file# Suitable for production environments#####################

# Bind the address, allow all IP access, and it is recommended to change the production environment to intranet IPbind 192.168.1.1

# Protect mode, it is recommended to enable (yes) to enhance securityprotected-mode yes

# Listen to the portport 6379

# TCP connection queue sizetcp-backlog 511

# Connection timeout (0 means no timeout)timeout 0

# TCP keep-alive time (seconds), it is recommended to set it to be larger to avoid the connection being accidentally disconnected by the firewall.tcp-keepalive 300

# Backend rundaemonize yes

# Process PID file pathpidfile /data/redis/redis_6379.pid

# Log level (debug | verbose | notice | warning)loglevel notice

# Log file path (null value indicates output to standard output)logfile "/data/app/redis/logs/"

# Number of databases (default 16, adjusted according to business needs)databases 16

# Show Redis startup logo (shutdown to reduce log interference)always-show-logo no

Basic network and process management

bind and protected-modeSpecified in the configuration filebind 192.168.1.1, only this IP is allowed to access the Redis service. For production environments, it is recommended to use intranet IP to limit access. Turn on at the same timeprotected-mode(Protected Mode), which can prevent unauthorized access.
Port and connection settingsuseport 6379Set the Redis listening port,tcp-backlogThe length of the connection queue is set.timeout 0Indicates that the idle connection is not automatically disconnected, andtcp-keepalive 300Maintain long connection activity and avoid intermediate devices (such as firewalls) being disconnected due to idle timeout.
Background operation and loggingpassdaemonize yesMake Redis run in the background and define the PID file path (pidfile) and log file path and log level. These settings help manage processes and troubleshoot problems.

RDB persistent configuration

Redis provides RDB persistence method, which can save memory data to disk regularly. The following optimizations are made for RDB in the configuration file:

#####################
# RDB persistence#####################
# Block writes when RDB fails to avoid data corruptionstop-writes-on-bgsave-error yes
# Enable RDB data compressionrdbcompression yes
# Enable RDB data verificationrdbchecksum yes
# RDB file namedbfilename 
# Close RDB to delete synchronous files to prevent mistaken deletionrdb-del-sync-files no
# RDB file storage directory, it is recommended to set it to SSD diskdir /data/app/redis/data/

Data integrity protectionusestop-writes-on-bgsave-error yes, Once an error occurs in RDB persistence, stop the write operation to prevent data inconsistency.
Data compression and verificationTurn onrdbcompressionandrdbchecksumIt can effectively reduce the size of RDB files and ensure data integrity through verification.
File storage directoryStore RDB files in/data/app/redis/data/, it is recommended to deploy on SSD for higher I/O performance.

Copy (master-slave synchronization) settings

In high availability architectures, master-slave replication is a common method. The following settings are made for the copy function in the configuration file:

#####################
# Copy (master-slave synchronization)#####################
# Allow replicas to still provide read-only service when disconnecting the main libraryreplica-serve-stale-data yes 
# Replica Node Read-only Modereplica-read-only yes
# Turn off disk-free synchronization (default is used)repl-diskless-sync no
# No disk synchronization delayrepl-diskless-sync-delay 5 
# Turn off disk-free loadingrepl-diskless-load disabled
# Keep the default TCP nodelay configurationrepl-disable-tcp-nodelay no
# Replica priority (the smaller the easier it is to become the main library)replica-priority 100

Replica service availabilityreplica-serve-stale-data yesIt allows the replica to continue to provide read-only services when it is disconnected from the main library to ensure that the business is not interrupted.
Synchronization methodDisk synchronization is used by default, through adjustmentrepl-diskless-sync-delayTo control the delay and maintain the stability of data transmission.
Master selection strategyConfigurationreplica-priority, the lower the value, the more likely it is to be elected as a new master library when the master library fails.

Memory management

Memory management is a core part of Redis performance, and flexible memory management strategies are provided in the configuration file:

#####################
# Memory Management#####################
# The default memory is not limited, and can be adjusted according to business needs.maxmemory 0
# If you do not evict data, you can change to allkeys-lrumaxmemory-policy noeviction
# Turn off lazy deletion to avoid extra CPU overheadlazyfree-lazy-eviction no 
# Turn off lazy expirationlazyfree-lazy-expire no 
# Turn off lazy deletionlazyfree-lazy-server-del no 
# Close the lazy copy cleaningreplica-lazy-flush no
# Close OOM Adjustmentoom-score-adj no 
# Enable only when OOM protectionoom-score-adj-values 0 200 800

Memory limits and policiesmaxmemory 0Indicates that the default unlimited memory is suitable for environments with sufficient memory resources; at the same time, it is setmaxmemory-policy noeviction, means that the data will not be evicted when the memory reaches the upper limit. In actual production, you can choose according to business needs such asallkeys-lruSuch as an expulsion strategy.
Lazy deletionTurning off various lazy delete functions (lazyfree-* configurations) can reduce CPU overhead, but may make the delete operation more synchronously executed and need to be traded down based on business scenarios.

AOF persistence

AOF (Append Only File) is another persistence solution for Redis that can provide higher data security. The AOF is set as follows in the configuration file:

#####################
# AOF Persistence#####################
# Enable AOF persistenceappendonly yes 
# AOF file nameappendfilename "" 
# Synchronize once per second, a compromise between performance and securityappendfsync everysec
# Whether to turn off synchronization during rewritingno-appendfsync-on-rewrite no
# The proportion of triggering AOF rewriteauto-aof-rewrite-percentage 100 
# Minimum size to trigger AOF rewriteauto-aof-rewrite-min-size 64mb
# Allow truncated AOF to loadaof-load-truncated yes
# AOF is compatible with RDB headers, reducing restart timeaof-use-rdb-preamble yes

Data security and performanceTurn onappendonly yesAnd useappendfsync everysec, strike a balance between data security and performance; synchronizing AOF files once a second can significantly reduce the risk of data loss.
Automatic rewriteConfigurationauto-aof-rewrite-percentageandauto-aof-rewrite-min-size, ensuring that the AOF file does not increase without limit, while using incremental rewrite to reduce performance losses during rewrite.
Compatibility and recoveryEnableaof-use-rdb-preambleThe RDB header data can be used to speed up loading speed and improve recovery efficiency during restart.

Performance optimization settings

In order to obtain better response speed in high concurrency environments, Redis also made a series of performance tuning in the configuration file:

#####################
# Performance optimization#####################
# Turn up the Hz frequency to increase the response speedhz 50
# Dynamic adjustment Hzdynamic-hz yes 
# AOF incremental synchronization during rewritingaof-rewrite-incremental-fsync yes
# RDB incremental synchronization during savingrdb-save-incremental-fsync yes 
# Enable jemalloc thread-optimized memory managementjemalloc-bg-thread yes

Hz frequencyThe default event processing frequency (hz) is raised to 50 times/sec, and dynamic adjustment is enabled to ensure that it can still respond to client requests quickly when load fluctuates.
Incremental synchronizationEnable incremental synchronization for AOF rewrite and RDB save can effectively reduce the pressure on disk I/O and improve overall performance.
Memory allocation optimizationEnable jemalloc background threads (jemalloc-bg-thread yes) Further optimize memory allocation and release, suitable for high concurrency scenarios.

Restrictions and monitoring

To detect problems in a timely manner and prevent unexpected situations, Redis provides a series of monitoring and restrictions:

#####################
# Restrictions and monitoring#####################
# Slow query threshold (microseconds)slowlog-log-slower-than 10000
# Slow query log maximum numberslowlog-max-len 128 
# Turn off delay monitoringlatency-monitor-threshold 0 
# Close key event notificationnotify-keyspace-events "" 
# No limit for ordinary clientsclient-output-buffer-limit normal 0 0 0
# Replica Node Limitclient-output-buffer-limit replica 256mb 64mb 60
# PubSub Limitationsclient-output-buffer-limit pubsub 32mb 8mb 60

Slow query logBy settingslowlog-log-slower-than 10000(in microseconds) to record commands with execution time of more than 10 milliseconds help to locate performance bottlenecks.
Client output bufferBuffer size limits are set for ordinary clients, replication nodes and PubSub modules respectively to prevent abnormal situations (such as client blocking) from causing memory surges.

Other parameters

Finally, some additional parameters are defined in the configuration file, such as RDB saving conditions and TCP-related parameters, to further refine the behavior of Redis:

#####################
#Other parameters#####################
# RDB Trigger Conditionsave 900 1 300 10 60 10000 
# TCP connection queue sizetcp-backlog 511
# TCP Keep-alive time (seconds)tcp-keepalive 300

These parameters can fine-tune the data storage frequency, network connection queue, etc. according to specific business scenarios, thereby achieving a balance between performance and reliability.

Complete configuration

#########################################
# Redis configuration optimization file# Suitable for production environments#########################################

# Bind the address, allow all IP access, and it is recommended to change the production environment to intranet IPbind 192.168.1.1

# Protect mode, it is recommended to enable (yes) to enhance securityprotected-mode yes

# Listen to the portport 6379

# TCP connection queue sizetcp-backlog 511

# Connection timeout (0 means no timeout)timeout 0

# TCP keep-alive time (seconds), it is recommended to set it to be larger to avoid the connection being accidentally disconnected by the firewall.tcp-keepalive 300

# Backend rundaemonize yes

# Process PID file pathpidfile /data/app/redis/redis_6379.pid

# Log level (debug | verbose | notice | warning)loglevel notice

# Log file path (null value indicates output to standard output)logfile "/data/redis/logs/"

# Number of databases (default 16, adjusted according to business needs)databases 16

# Show Redis startup logo (shutdown to reduce log interference)always-show-logo no

#########################################
# RDB persistence#########################################
# Block writes when RDB fails to avoid data corruptionstop-writes-on-bgsave-error yes
# Enable RDB data compressionrdbcompression yes
# Enable RDB data verificationrdbchecksum yes
# RDB file namedbfilename 
# Close RDB to delete synchronous files to prevent mistaken deletionrdb-del-sync-files no
# RDB file storage directory, it is recommended to set it to SSD diskdir /data/app/redis/data/

#########################################
# Copy (master-slave synchronization)#########################################
# Allow replicas to still provide read-only service when disconnecting the main libraryreplica-serve-stale-data yes 
# Replica Node Read-only Modereplica-read-only yes
# Turn off disk-free synchronization (default is used)repl-diskless-sync no
# No disk synchronization delayrepl-diskless-sync-delay 5 
# Turn off disk-free loadingrepl-diskless-load disabled
# Keep the default TCP nodelay configurationrepl-disable-tcp-nodelay no
# Replica priority (the smaller the easier it is to become the main library)replica-priority 100 

#########################################
# Memory Management#########################################
# The default memory is not limited, and can be adjusted according to business needs.maxmemory 0
# If you do not evict data, you can change to allkeys-lrumaxmemory-policy noeviction
# Turn off lazy deletion to avoid extra CPU overheadlazyfree-lazy-eviction no 
# Turn off lazy expirationlazyfree-lazy-expire no 
# Turn off lazy deletionlazyfree-lazy-server-del no 
 # Close the lazy copy cleaningreplica-lazy-flush no
 # Close OOM Adjustmentoom-score-adj no 
# Enable only when OOM protectionoom-score-adj-values 0 200 800

#########################################
# AOF Persistence#########################################
# Enable AOF persistenceappendonly yes 
# AOF file nameappendfilename "" 
# Synchronize once per second, a compromise between performance and securityappendfsync everysec
# Whether to turn off synchronization during rewritingno-appendfsync-on-rewrite no
# The proportion of triggering AOF rewriteauto-aof-rewrite-percentage 100 
# Minimum size to trigger AOF rewriteauto-aof-rewrite-min-size 64mb
# Allow truncated AOF to loadaof-load-truncated yes
 # AOF is compatible with RDB headers, reducing restart timeaof-use-rdb-preamble yes 
#########################################
# Performance optimization#########################################
# Turn up the Hz frequency to increase the response speedhz 50
# Dynamic adjustment Hzdynamic-hz yes 
# AOF incremental synchronization during rewritingaof-rewrite-incremental-fsync yes
# RDB incremental synchronization during savingrdb-save-incremental-fsync yes 
 # Enable jemalloc thread-optimized memory managementjemalloc-bg-thread yes

#########################################
# Restrictions and monitoring#########################################
# Slow query threshold (microseconds)slowlog-log-slower-than 10000
# Slow query log maximum numberslowlog-max-len 128 
# Turn off delay monitoringlatency-monitor-threshold 0 
# Close key event notificationnotify-keyspace-events "" 
# No limit for ordinary clientsclient-output-buffer-limit normal 0 0 0
# Replica Node Limitclient-output-buffer-limit replica 256mb 64mb 60
# PubSub Limitationsclient-output-buffer-limit pubsub 32mb 8mb 60

#########################################
#Other parameters#########################################
# RDB Trigger Conditionsave 900 1 300 10 60 10000 
# TCP connection queue sizetcp-backlog 511
# TCP Keep-alive time (seconds)tcp-keepalive 300

Summarize

This blog post analyzes a Redis configuration file optimized for production environments in detail. By reasonably setting network parameters, persistence policies, replication mechanisms, memory management and performance tuning, Redis can maintain stable and efficient operation in high concurrency and large data volume scenarios.

During actual deployment, various parameters should be adjusted appropriately based on their own hardware conditions and business characteristics. For environmental security, data reliability and performance optimization, it is recommended to refer to official documents and community best practices to continuously improve and tune configurations.

This is the article about the stand-alone configuration of Redis6.2.6 production environment. For more related Redis configuration content, please search for my previous articles or continue browsing the related articles below. I hope everyone will support me in the future!