Misuse of the server and proper use

The server is a critical core device, ensuring that the network server is capable of high performance, stable and continuous work has always been the user's biggest concern. However, while paying attention to this problem, we found that many users did not configure their own servers correctly, so that the server did not work in the best state. Usually the common misunderstandings of server configuration are shown in the following aspects.

Misunderstanding on server usage

One of the misunderstandings: Server with redundancy instead of

Many high-performance servers provide array functionality, but because of users I don't know, I only buy one hard drive, no data redundancy, lost security and performance optimization for storage

Misunderstanding 2: High-end server uses low configuration scheme

User-purchased high-end The server itself can meet high performance requirements, but it is equipped with low-speed, small-capacity hard disk and small-capacity memory, which causes the overall performance of the server to be greatly reduced.

Misunderstanding 3: Do not understand the server Resource bottleneck caused by performance bottleneck

Some users have too one-sided understanding of the server, unilaterally recognize the importance of certain components, and do their best to invest exclusively, while ignoring the optimization and upgrade of other components. Work, resulting in the performance of these components has not been played out. It can be seen from the above analysis that the operation status of many servers is not optimal, either because the user does not fully understand the functions provided by the server or the misunderstandings in the use and configuration. According to statistics, 80% of the servers in the industry are not optimized, 90% of servers do not have regular system performance monitoring, 95% of servers do not have comprehensive data redundancy security measures, and nearly half of servers do not use data backup solutions. . These servers are actually in a sub-health state, which is embodied in the following: no critical redundancy for power, fans, hard disks, controllers, cables, network cards, CPUs, etc., which leads to system security. Reduced performance; use low-speed, compatible device components, unreasonable configuration of memory, CPU, hard disk controller, etc., resulting in performance degradation; do not use any network server management software and hardware, when the failure occurs, the system is down, management There are serious flaws.

How to use the server correctly

In order to improve the health of the server, Yuehai Yangbo specializes in some of the current shortcomings and puts forward feasible suggestions:
for the hard disk The storage section adds redundant hard drives and array control cards to provide data redundancy and dramatically increase system IO performance.
Add redundant CPUs to servers, use SMP (Symmetric Multiprocessor) technology to improve system performance, and increase central processing redundancy.
Add redundant network cards to improve the IO performance of the network. When a network card fails, the server will not be disconnected from the network.
Add redundant power modules to the server to improve the power supply capacity of the server. When a power module has a problem, the system will not be shut down due to power interruption.
Increase memory for servers, meet operating system requirements and increasing optimization and application requirements, and improve server performance.

In addition, you need to balance the overall performance of the server to avoid performance bottlenecks and security risks. From CPU processing power, memory size, data redundancy and data storage IO capabilities, network IO performance, power supply capability, fan cooling capacity, system fault alarm capability, and power failure repair capabilities are optimized. Work, such as: increase the hard disk, array card, increase the cache of the array card, select the hot-swappable hard disk bracket, use multiple channels of the array card, select the most suitable array level to meet different read and write performance to optimize Storage subsystem.

According to the operating system used, the number of users, the scope of application, the number of CPUs used, the minimum memory capacity is determined, and the remote control card is diagnosed online to diagnose faults during memory operation.

According to the processing power required by the system, the system's redundancy requirements for the CPU, the number of users, and the scope of application to determine the number of CPUs used, use the operating system performance monitoring software and network management software to detect the CPU occupancy. , decide the amount to be increased or decreased.

Improve network IO performance of server NICs by using NIC redundancy technologies such as AFT (Network Card Redundancy), ALB (Network Load Balancing), and FEC (Fast Ethernet Channel).

Add redundant power supply modules to effectively guarantee the server power supply, prevent system downtime due to damage of a single power module, increase or decrease the redundancy of the fan to ensure the system cooling effect, and prevent the server from overheating. malfunction.

In short, carefully check the existing bottlenecks and defects, tailor each part to optimize the obstacles, and fully guarantee the input-output ratio, which can make you use properly. Your server resources, avoiding the misunderstanding of entering the server.