A few things you need to know before buying a server (on)

For enterprise users who want to choose a suitable server, they must first have a correct understanding of server usage. When making server selection, it should be considered according to the following three aspects.

1. Understand your application needs

refers to the main application of the entire system. Specifically, it is the number of users supported by the server, the type of user, the amount of data processed, and so on. Different application softwares work differently, and the requirements for server matching are very different. Common applications can be divided into file services, Web services, general applications and databases.

2. Usability

The server is the core of the entire network, not only to meet the needs of network applications in terms of performance, but also to provide services to network customers without interruption. In fact, the reliable operation of the server is the basis for the stable function of the entire system.

3. Optimized configuration

Server types, such as low-end, mid-range and high-end categories, only determine the maximum number of users that the server can support. But to use the server well, you need to optimize the configuration to get the best performance at the lowest cost.

Common Application Analysis

In the SMB environment, common applications can be summarized as the following, and their requirements for the server are different. For the convenience of description, the server is divided into four functional modules, namely CPU, memory, disk subsystem and network subsystem.

1. File Service

This is the most basic application service. The server is equivalent to a large warehouse of information systems to ensure fast data transfer between users and server disk subsystems. Among the various subsystems of the server, the system subsystem has the greatest impact on system performance, followed by the disk subsystem, and then the memory capacity, and the CPU requirements are generally not high

2. Database service < Br>

The most demanding applications for all aspects of the system (except the network subsystem), such as finance, inventory, and personnel management applications. A high performance CPU and a fast disk subsystem are required to satisfy a large number of random I/O requests and data transfers. The server bottlenecks are: memory, disk subsystem, and CPU.

3. Mail Service

Play the role of email router and warehouse. The server bottlenecks are: network subsystem, memory, disk subsystem, and CPU.

4.Web Services

The performance of the server is determined by the content of the website. If the Web site is static, the system bottlenecks are: network subsystem and memory. If the web server is mainly engaged in intensive computing (such as dynamically generating web pages), the system bottlenecks are: memory, CPU, disk subsystem, and network subsystem.

5. Multimedia Services

Responsible for media control and media streaming functions on the network, I /O throughput has a critical impact on server performance. The bottlenecks of the video server are: network subsystem, disk subsystem, and memory. Audio services have very low requirements for server hardware configuration, and today's server subsystems generally do not become bottlenecks.

6. Terminal Services

Execute various applications and pass the results to the user, all the load is added to the server. The system bottlenecks are usually: memory, CPU, network subsystem.

7. Primary Domain Controller

The primary domain controller is the management center for networks, users, and computers, and is responsible for providing a secure network working environment. The primary domain controller not only responds to the user's login requirements, but also synchronizes and backs up user accounts, WINS and DHCP databases between servers. In addition, the primary domain controller also performs DNS services. The system bottleneck is the network subsystem and memory.

Impact of Availability

A server that crashes frequently is unbearable, and the resulting loss is not only a waste of time, but also a multi-day workload. . More and more people are now aware of the importance of system availability.

Availability is usually measured by the system's theoretical uptime and percentage of actual usage. For example, we say that a system provides 99% availability in a 24x7 environment, which means that it may be down for 88 hours a year, which is unacceptable to most users. 99.999% availability guarantees a system downtime of 5.25 minutes a year, but the price of this system is very expensive.

The availability of the server depends mainly on two aspects: one is the quality of the server itself, which is embodied in the professional design of the server manufacturer, strict quality control and long-term verification of the market; Protection measures taken by the loss component, such as: NIC redundancy technology, disk array technology, power redundancy technology, dual-machine or cluster solution to ensure online redundancy of the network, disk, power supply and even the entire host.

In low-end servers, the following measures are usually taken to improve the availability of a single machine.

1.IDE RAID

Provides data redundancy through inexpensive disk arrays. Disk failure is the main cause of server hardware failure, with a failure rate of up to 52%. The harm of data loss is also amazing, resulting in a lot of time and manpower waste. Currently, IDE RAID can implement RAID-0, RAID-1, and RAID-0+1 in three ways. Among them, RAID-0 does not have data redundancy, but it can significantly improve the performance of the disk subsystem.

2.ECC Technology

It is possible to check two memory errors and correct one bit error to ensure high reliability of data in the memory and cache.

3. Server-specific power supply

can ensure that the system has a clean power environment, reducing the occurrence of various hidden faults, and inferior power supply is easy to cause all kinds of weird faults, such as in the circuit. The high frequency crosstalk will cause the system to crash frequently, and the low frequency oscillation will burn the electronic components invisible. Such faults also increase the difficulty of maintenance.

4.Additional measures

If the design of the air filter, the heat dissipation of multiple fans (some servers also have the function of automatically adjusting the fan speed), it can help the server in the general environment. Stable operation.