Server

Server refers to a management resources, and provides services for users of computer software, usually consist of a file server, database server and application server. Running over the computer software or computer systems also called server. Relative to ordinary PC speaking, server in stability, security, performance and so on, and therefore require higher CPU, chipsets, memory, and a magnetic disc system, network hardware and ordinary PC differs somewhat.

Server parsed

Server as network nodes, storage, processing network in 80% of the data, information, so also called network soul. Do an image of the parable: the server is like the post office switches, and microcomputer, notebook, PDA, cell phone fixed or mobile network terminal, like family, scattered in various workplaces, public place of the place such as telephones. Us with the outside the daily life, work of the telephone exchanges, communication, must pass a switch to reach target telephone, The same is true with network terminal equipment such as family, enterprises in the microcomputer online, access to the information, communicates with the outside world, entertainment and so on, also must pass a server, so also say, is the server "organization" and "leading" the equipment. It is a kind of network for client computer providing various kinds of service of high availability computer, it's in the network operating system under the control of will connected to it hard disks, tapes, printer, Modem and various special communication equipment provided to the client on site sharing network, can also for network users provide centralized calculation, information published and data management services. Its high-performance mainly reflects in high speed of operation ability, long and reliable operation, strong external data throughput capacity, etc.

The server's structure and microcomputer basic similar, have processor, hard disk, memory, and system bus, etc., which is aimed at a specific network application tailored, thus server and microcomputer in handling capacity, stability, reliability, security, expansibility, managerial aspects to have difference is very big. Especially along with the progress of information technology, network effect is more and more obvious, to oneself the information system of data processing power, such as security requirements are increasingly high, if you are in the process of conducting e-commerce by hackers steal password, losses key business data, If you in ATM's not normal to access, you should consider in these equipment system behind the commander - the server, not blame work personnel's quality and other objective conditions restrictions.

Server classification

According to the architecture of the handout

At present, according to the architecture will distinguish, server mainly divided into two classes:

Non x86 server: including mainframe, minicomputer and UNIX server, they are used as RISC instruction set) or EPIC (parallel instruction code) processor, and mainly USES the UNIX operating system and other special servers, reduced instruction set processor are mainly on IBM's POWER and PowerPC processor, from Fujitsu cooperation with r&d SPARC processor, EPIC processor is mainly HP and AnTeng processor of Intel cooperation development etc. This server is expensive, system closed, but good stability and performance is strong, basically use in finance, telecommunications and other large enterprise's core system.

X86 server: also called (CISC complex instruction set) architecture server, namely usually speak PC server, which is based on PC System structure, the use of Intel or other compatible x86 instruction set processor chips and Windows operating System of the server, such as IBM's views on the x server, HP series of Proliant series server etc. Cheap price, good compatibility, poor stability, unsafe, basically use in small and medium-sized enterprises and non-critical business.

From the current network development situation look, by the small, chocolate, and stable "feature of x86 framework PC server received more widely. From a theoretical definition, server is in network environment of high performance computer on the network, it protected reliably against detective other computers (client) submitted to the service request, and provide the corresponding services. Therefore, the server must undertake service and security has the quality of service ability.

But this to explain still appear relatively abstruse fuzzy, actually server and the function of Personal computers are similar, all is help mankind to process information tools, just two orientation is different, Personal computers (referred to as the Personal Computer, PC) is for our Personal function, need and design, and the server is to meet many users in its at the same time on data processing and design. While many people how to use same server? This can only be done through network interconnection, to help achieve the common use of purpose.

Let the server functions, the server can be used to build a web service (see our normal Internet web page data is stored in the server offer people visit), mail service (we send all E-mail all need to pass the server processing, sending and receiving), file sharing & printing Shared services, database services, etc. And this all applications have one common feature, they face all is not a person, but many people, while processing is a range of data. So the server and network is inseparable. Can say, there is no left the network server. The server is born for providing services, only on the network environment it then have the value of existence. And personal computers can completely in the offline complete master data processing tasks.

According to the application level division

According to the application level classification is also commonly known as "press server class to differentiate" or "according to network scale" points, the most common is server of a kind of method to partition, it mainly based on the server in the network application level (or server class to) divided. Care must be taken citcc server class and not by server CPU frequency discretion to differentiate, but by the whole server comprehensive properties, especially the adoption of some servers special technology to measure. According to this classification method, the server can be divided into: entry-level server, workgroup level server, BuMenJi server, enterprise-level server.

1, entry-level server

This kind of server is the most basic category server, is also the most cheap server. As the PC technology is increasing day by day, now many entry-level server with the PC configuration, so now almost there are some people think entry-level server and "PC server" equivalent. This kind of server contains server features and not many, usually only possess the following several aspects features:

· some basic hardware redundancy, such as the hard drive, power supply, fan etc, but not a must;

· usually adopts SCSI hard disk interface, also now have adopted SATA serial interface,

· some components support hot plugging dial, such as hard disk and memory, etc, these also is not a must;

Only a CPU, usually, but not absolutely, such as from the entry-level server some can support to 2 processor,

· memory capacity also is not very large, be in commonly 1GB, but usually by within.ecc correction technology with the server special memory.

This kind of server mainly USES the Windows or NetWare network operating system, can fully satisfy office type of small and medium-sized network users file sharing, data processing, Internet access and simple database application requirements. This server with general PC are very similar, there are many small companies with a simply high-performance brand PC as a server, so this kind of server whether in performance, it is the price with a high-performance PC PinPaiJi yielding around, such as DELL the latest PowerEdge4000 SC price only 5808 yuan, HP has similar configuration and price entry-level server.

Entry-level server even terminals are more limited (usually 20 sets control), moreover in stability, expansibility and fault-tolerant redundancy performance is poorer, applies only to no large database data exchange, daily work network traffic is not big, need not for a long time, uninterruptedly boot of small businesses. But to explain a bit is currently some relatively large server development, production manufacturers behind we will speak enterprise-class server also differentiate few class, one of the most cheap an enterprise-class server class is called "entry-level enterprise-level server," said by here the entry-level and not with our the foregoing "entry-level" with the same meaning, but this classification or less. Also the fact that this server usually adopts the intel-based dedicated server CPU chip, is based on the Intel architecture (known as "IA structure"), which, of course, is not a rigid standard rules, but because the server application levels needs and price limits.

2, workgroup server

Workgroup server is a than entry-level high a level of the server, but still belongs to such low-grade server. From this name also can see, it only can connect a work group (50 Taiwan controlled) so many users, the network is smaller in scale server stability is not like below we will speak enterprise-class server as high application environment, certainly in other performance requirements are correspondingly lower. Workgroup server has the following aspects of the main features:

· usually only supports single or double CPU structure of the application server (but not absolute, especially the workgroup server from there can support as many as four processors workgroup server, of course, this type of server price also something different);

· can support high-capacity ECC memory and enhance server management function of SM bus,

· function more comprehensively and manageability is strong, and easy maintenance,

The Intel server, using CPU and Windows/NetWare network operating system, but also a part is to adopt the UNIX operating system; series

Internet users can satisfy the small and medium-sized, data processing, file sharing, Internet access and simple database application requirements.

Workgroup server is entry-level server for performance was improved, and the function improved, have certain scalability, but fault-tolerant and redundancy properties are still not perfect, also cannot satisfy a large database system application, but the price is more expensive than the former, generally equivalent to 2 ~ 3 sets of high performance PC PinPaiJi total price.

3, BuMenJi server

This kind of server is belong to the group stages, midrange server is commonly support twin CPU above symmetric processor structure, a comparatively complete hardware configuration, such as disk array, storage brackets etc. BuMenJi server that the biggest characteristic of, besides having workgroup server all servers characteristics outside, still a lot of integrated circuit monitoring and management, have overall server management ability, can monitor such as temperature, voltage, fan, box, such status parameters, combined with standard server management software, make management personnel to understand the server work conditions. In the meantime, most BuMenJi server has good system extensibility, can meet the users in turnover increased rapidly when can seasonable online upgrade system, fully protect the user's investment. It is the enterprise network dispersed each basic data acquisition unit and the top data center keep smooth connecting the necessary link, generally for a medium-sized enterprise's first choice, can also be used to finance, telecommunications, etc.

BuMenJi server usually adopts the IBM, from the development of their respective and HP CPU chip, such chips are generally RISC structure, adoption of the operating system is usually the UNIX operating system, now of the series BuMenJi server in LINUX can be widely used. Before BuMenJi server can production manufacturer usually only the IBM, COMPAQ, HP, from (now has merged into HP) so a few home, but now with some other server vendors development technology improvement, can now development, production BuMenJi server vendors have more many. Home also has several have the strength, such as lenovo, shuguang, tide and so on. Of course because there an industry standard stipulate what kind of server configuration can hardly is BuMenJi server, so now also have many strength is not strong enterprise also says it has BuMenJi server, but its product configuration is basically with entry-level server makes no difference, the user must pay attention.

BuMenJi server can connect 100 or so computer users, suitable for processing speed and reliability high some of small and medium-sized enterprises network and its hardware configuration is relative taller, its reliability level higher than the workgroup server some, of course, the price is higher also (usually for about 5 sets of high-performance PC price combined). Because this kind of server needs to install more parts, so usually larger, use machines chassis of the cabinet.

4, enterprise-level server

Enterprise-level server belongs to high-grade server ranks, because of this, can produce this server enterprise also is not much, but also because there is no industry standard hardware provisions to achieve what level of hongdu server, so now also saw many this does not have the development, the production of hongdu server level of business claim the enterprise-level server. Enterprise-level server at least is USES 4 above CPU symmetric processor structure, some as high as dozens of. Also general also have independent double PCI channel and memory expansion board design, has the high memory bandwidth, high-capacity hot plug and play hard disk and hot swaps power, super data processing and clustered property, etc. This kind of enterprise-level server box is bigger, generally for machine cabinet, some still by several cabinet to form, like mainframe as well. Enterprise-level server product besides have BuMenJi server all server properties outside, the biggest characteristic is that it also has high tolerance, excellent expansibility and fault forecasting warning function, online diagnosis and RAM, PCI, CPU has hot plugging dial the performance. Some enterprise-level server also introduced large computer of many excellent properties, such as IBM and from enterprise-class server. This kind of server adopted the chips are also a few large server development, production manufacturers develop their own unique CPU chip, adoption of UNIX operating system is also commonly (Solaris) or LINUX. At present in the global scope internal energy production upscale enterprise-level server vendors also only IBM, HP, from so several, the majority of domestic and foreign manufacturers enterprise-class server can be medium and low grade enterprise servers. Enterprise-level server for running in need to handle large data, high processing speed and reliability requirements of high finance, securities, transportation, post and telecommunications, communication or large enterprise. Enterprise-level server for networked computer in hundreds of above, processing speed and data safety requirements of very high large network. Enterprise-level server hardware configuration is highest, system reliability also is the strongest.

The server hardware

Actually sweetie-pie server system hardware structure and normally we contact with computer have many similarities, mainly the hardware constitution still contains the following several main parts: CPU, memory, and chipset, the I/O bus, I/O devices, power, box, and related software. It also became we choose and buy a server when its main concern indicators.

The whole server system is like a man, the processor is the server's brain, and various bus is like the nerve, systemic muscle distribution chipset like spinal cord, and I/O devices like through the nervous system dominate people's hand, eyes, ears and mouth, And power systems like blood circulation system, it will energy delivery to all parts of the body.

For a server speaking, server performance design goal is how to balance the performance of each part, make whole system performance to achieve optimal. If a server has per 1,000 service request processing ability, but nic can only accept 200 request, and hard disk can only afford 150, and various bus load ability can only assume 100 request that this server must processing power is only 100 request/SEC, more than 80% of the processor computing power wasted.

So to design a good server is through balance the final purpose of various aspects of the performance, making each part cooperate proper, and can give full play to ability. What can we learn from these aspects to measure the server whether has reached its design purpose, R: Reliability; Reliability - A: Availability, usability, S: Scalability; Scalability -- U: application - ease-of-use, M: Manageability - Manageability, namely the server RASUM criteria.

As the server in the network to provide service, so the quality of service to undertake various application network computing environment is very important, assume this service of computer hardware must be able to guarantee the quality of service. This service the first to have certain capacity, can response unit time reasonable quantity of server request, and the service of individual service request response time to try to fast, and this service in time required range has always been there.

If a WEB server only in 1 minute to handle the 1 page request, 1 other than requests must be queuing up to, and this one request must be 3 minutes to deal with, and the WEB server in one hour ago can access to, but after an hour but didn't connect, the WEB server in today's Internet computing environment is unthinkable.

Now the WEB server must be able to handle thousands of access, while each visit response time should be short, and the WEB server can't stop, otherwise this WEB server will result in the loss of access the user.

To achieve the above requirements, as the server hardware must have the following characteristics: performance, in unit time server can handle a considerable amount of server request and ensure each service response time, Reliability, making the server can not stop, Scalability, make sure that the server can as the number of users increase continuously improve performance. So we say cannot put an average PC as servers to use, because, PC far less than the above requirements. So we in the server's conceptual and plus a little is the server must have bear service and guarantee the quality of service ability. This distinguishes low-cost servers and PC differences of the major aspects.

In the information system, a server is mainly used in database and Web services, and PC mainly applies to the desktop computing and Internet terminals, design essential starting point of difference determines the server should have more reliable than PC continued operation ability, more powerful storage capacity, and network communication ability, rapider fault recovery function and broader patulous space, meanwhile, the application of data fairly sensitive to still required server provides data backup function. And PC is more attention in the design of human-machine interface usability, image and 3D processing power and other multimedia performance.