tuned is a dynamic tuning scheme introduced by Red Hat that allows users to adopt different tuning options over different time periods. Due to the form of service process, it can be very convenient and crontab combination! Tuned is a daemon that monitors and collects data on the usage of individual system components, and can use that information to dynamically adjust system settings as needed. It reacts to changes in CPU and network usage and adjusts settings to improve the performance of the active device or reduce the power consumption of inactive devices.
With its tool ktune combined with the Tuned-adm tool, it provides a number of preconfigured tuning analyses to improve performance and reduce energy consumption in a number of specific use cases. Editing these configurations or creating new configurations can create a system-specific performance solution.
View tuning scenarios for different scenarios in the System Tuned-adm list
Use some kind of profile Tuned-adm profile Server-powersave
A description of some profiles:
Default
The default power saving configuration. This is the most basic node configuration. It only enables disk and CPU plug-ins. Note: This is different from shutting down the TUNED-ADM, which disables both tuned and Ktune.
Latency-performance
Server configuration for typical deferred performance debugging. It disables the tuned and Ktune energy-saving mechanisms. Cpuspeed mode changed to performance. The I/O riser for each device is changed to deadline. For the power management quality of the service, set the cpu_dma_latency to 0.
Throughput-performance
Server-side write for typical throughput performance tuning. This is recommended if the system does not have enterprise-class storage. It is the same as latency-performance, just:
Set the Kernel.sched_min_granularity_ns (Scheduler minimum priority time interval) to 10 milliseconds,
Set the Kernel.sched_wakeup_granularity_ns (scheduler wake-up interval) to 15 milliseconds.
Set the Vm.dirty_ratio (virtual machine Dirty data scale) to 40% and enable transparent oversized pages.
Enterprise-storage
This profile is recommended for most enterprise server configurations, including battery backup control program cache protection and managing disk caches. It is similar to the throughput-performance configuration, except that the file system is to be re-mounted using barrier=0.
Virtual-guest
This profile is recommended for most enterprise server configurations, including battery backup control program cache protection and managing disk caches. It is similar to throughput-performance, just:
Set the ReadAhead value to 4x without using the Root/boot file system that barrier=0 re-mount.
Virtual-host
Depending on the enterprise-storage configuration, virtual-host can also reduce the amount of virtual memory that can be replaced and enable more collection Dirty page writeback. You can find this configuration in Red Hat Enterprise Linux 6.3 and later, and it is recommended to use this configuration in virtualized hosts, including KVM and Red Hat Enterprise Linux virtualization hosts.
Configuration file:/etc/tune-profiles/different profiles exist in directory form!
Create a tuning solution for your application:
Cd/etc/tune-profiles;
Cp-r Enterprise-storage/my-server;
CD my-server/
Modify the response profile to add tuning parameters
Kdump is a kexec-based Linux kernel crash capture mechanism that saves kernel memory images before a crash, and the programmer analyzes the file to find out the cause of the kernel crash for system improvement. http://www.ibm.com/developerworks/cn/linux/l-cn-kdump1/
auditd-linux Server Security Audit tool https://linux.cn/article-4907-1.html http://blog.chinaunix.net/ Uid-17238776-id-4904716.html
irqbalance Theoretically:
Enable the Irqbalance service to improve performance and reduce power consumption.
Irqbalance is used to optimize interrupt allocation, which automatically collects system data to analyze usage patterns and puts the working state into performance mode or Power-save mode based on system load conditions.
When in performance mode, Irqbalance distributes interrupts as evenly as possible to each CPU core to take full advantage of CPU multicore and improve performance.
When in Power-save mode, the Irqbalance allocates interrupts to the first CPU to ensure the rest of the idle CPU's sleep time and reduce power consumption.
However, in practice the CPU is often affected by the use of equalization, it is recommended to shut down in the server environment. http://blog.yufeng.info/archives/2422
CentOS 7 System Services AUDITD Kdump tuned irqbalance