1. What is a server?

　　Server is a kind of computer on the network which provides various services for customer sites. Under the control of network operating system, it will provide the hard disk, tape, printer, Modem and expensive special communication equipment connected with it to the customer sites on the network to share, and also can provide centralized computing and database for network users.Management and other services.  

● The role of network server:
　　A． Run the network operating system. Through the network operating system to control and coordinate the operation of network workstations, processing and responding to various network operation requests from each workstation at the same time.  
　　B． Storage and management of hardware and software shared resources in the network, such as databases, files, applications, printers and other resources.  
　　C． The network administrator monitors, controls and adjusts the activities of the workstations on the network server.  

　　Structurally, servers are moving from RISC servers to IA servers, especially in small and medium-sized networks.

2.Server related technology

● hot plug

　　○ Hot Swap，Also known as heat exchange technology, hot swap technology. Allows the server to replace the hot plug devices such as the failure hard disk without shutting down.  
　　○ Combined with RAID technology, hot switching technology can make the server replace the fault hard disk in the state of shutdown, and automatically restore the data on the original disk, greatly improving the fault tolerance of the server system.  
　　○ Hard disk hot swap has two ways:
　　　A． Hot plug hard disk box with ordinary SCSI hard disk is used in disk array.  
　　　B． A dedicated hard disk with hot plug capability is the standard configuration of high performance servers.  
　　○ Hot plug technology will be developed in the future to hot plug power, hot plug PCI card and so on.  

● Hard disk interface technology

　　IDE： (Intergraded drive electronics) Now the mainstream hard disk interface used by PC.  
　　SCSI：(Small Computer System Interface) Minicomputer system interface. SCSI technology originated from minicomputer, and has been ported to PC servers and high-end PC machines. Compared with IDE interface, SCSI interface has the following performance advantages:
　　a. Intelligent interface independent of hardware devices: lighten the burden of CPU.  
　　b. Multiple I/O parallel operations: therefore, SCSI devices have fast transmission speed.  
　　c. The number of peripherals that can be connected is large: it can extend multiple peripherals (such as hard drives, tape drives, etc.).  

　　When the number of network users accessing the server at the same time is large, the I / O performance of the system using SCSI hard disk is obviously better than that of the system using IDE hard disk.  
　　SCSIThe bus supports fast data transmission. Different SCSI devices usually have 8 or 16 bit SCSI transmission buses. In multitasking operating systems, such as Windows NT, multiple SCSI devices can be activated at the same time. SCSI adapters usually use the DMA of the host (directly).The memory access channel transmits data to memory. This means that without the help of host CPU, the SCSI adapter can transfer data to memory. In order to manage data flow, every SCSI device (including adapter card) has an identity number. Usually, the identity number of the SCSI adapter is taken.Set to 7, and the identity numbers of the remaining devices are 0 to 6.  
　　Most PC based SCSI buses use single ended transceivers to send and receive signals. However, as the transmission rate increases and the cable lengthen, the signal will distort. Differential transceivers can be added to SCSI devices to maximize the bus length and ensure that the signal is not distorted. Differential transmissionThe device uses two wires to transmit signals. The second line is a copy of the signal pulse. Once the signal reaches its destination, the circuit compares the pulses of the two lines and generates a correct copy of the original signal.  
　　A new differential transceiver, LVD (Low Voltage Differential Transceiver), can increase the bus length and provide higher reliability and transmission rate. LVD can connect 15 devices, and the maximum bus length can reach 12 meters.  

SCSI series currently used:

Narrow Wide		Wide
Interface	transmission speed	Interface	transmission speed
Fast Fast SCSI	10 MB/S	Fast Wide SCSI	20MB/S
Ultra Ultra SCSI	20MB/S	Ultra Wide SCSI	40MB/S
Ultra2 Ultra2   SCSI	40MB/S	Ultra2 Wide SCSI	80MB/S
/		Ultra 3	160MB/S

SCSIDifference from IDE 

　　○ IDEThe way it works requires the full involvement of the CPU; in Windows 95/NT multitasking operating systems, this naturally leads to a dramatic slowdown in system response. The SCSI interface, on the other hand, controls the read-write operation of the data entirely through a separate high-speed SCSI card, so the CPU does not have toWasting time and waiting can obviously improve the overall performance of the system.  

　　○ SCSIGenerally, there are two IDE channels in each IDE system, connecting four IDE devices in total. SCSI interface can connect 7 to 15 devices, much more than IDE, and the connection cable is much longer than IDE.  

　　Although the cost of SCSI equipment is higher, compared with IDE, SCSI is more stable, durable and reliable.

● RAIDTechnology

　　○ RAID：（Redundant Array of Inexpensive Disk）Redundant array of inexpensive disks. As disk access speed can not keep up with the development of CPU processing speed, it becomes a bottleneck to improve server I / O capability. RAID technology uses disk segmentation, disk mirroring, data redundancy technology to improve disk access speed, while providing disk data backup, improve the systemReliability.  

　　○ Disk Striping: Data is read and written to multiple disks in “segments” in turn, and multiple disks are equivalent to simultaneous operations, greatly improving the speed of access.  

　　○ Disk Mirroring: Two disks are controlled by a single controller, and the same data is read and written at the same time. The data is backed up 100%.  

　　○ Data redundancy technology: data read and write to do a check, check data in a compact format stored on disk, can be used for error correction and recovery of data.  
　　○ RAIDThere are several commonly used technologies:

RAIDlevel	describe	technology	speed	Fault-tolerant ability
RAID 0	Disk segmentation	No check data	Disk parallel I/O, access speed increased the largest.	Data without backup
RAID 1	Disk mirroring	No check data	The speed of reading data is improved.	Data 100% backup (waste)
RAID 2	Disk segmentation + Hamming code data correction	/	No improvement	Allow single disk error
RAID 3	Disk segmentation + parity check	Special calibration data panel	Disk parallel I/O, the speed increases greatly.	Allow a single disk error, except for the check disk.
RAID 4	Disk segmentation + parity check	Asynchronous special check data panel	Disk parallel I/O, the speed increases greatly.	Allow a single disk error, except for the check disk.
RAID 5	Disk segmentation + parity check	Check data distribution is stored in multiple disk.	Disk parallel I/O is much faster, slower than RAID 0.	Allow a single disk error, no matter which disk.

　 
After RAID 5 is done, the system can still run after any disk fails, and the data on the failed disk can be calculated by checking data on other disks (at this time the speed is slower). If there is a backup disk in the disk system, the data is automatically restored to the backup disk. If you have hot plug hard diskIn the boot state, the hard disk can be replaced and the data will be automatically restored to the new hard disk. In these processes, the system did not stop running.

● SMPTechnology brief

　○ SMP：Symmetric Multiprocessing . Symmetric multiprocessing. Refers to a set of processors (multiple CPU) on a computer. Multiprocessing refers to the coordination of multiple processors in a computer by sharing the same storage area. True multiprocessing requires that each CPU in the system access the same physical memory. This means more CP.U must be able to use the same system bus or system switching mode.  
　　Operating system support for multiprocessing architecture is closely related to its core, which involves two basic sequence algorithms for supporting multiprocessing: symmetric and asymmetric processing. In asymmetric processing, CPU has different tasks; in symmetric processing, each CPU can perform any task. The SMP system will passProcessing load is distributed to each idle CPU to enhance performance. In handling distributed or executing threads, the functions of each CPU are the same. They share the memory and bus structure, and the system distributes the processing task queue symmetrically on multiple CPUs, thus greatly improving the data processing capacity of the system.  

　○ Symmetric multiprocessing first shows high performance in network management, which is attributed to the strong processing capacity of SMP system and the rise of SMP operating system. SMP supported network operating system: Novel Netware, SCO UNIX, Microsoft WindowS NT and so on.  

　○ SMPTechnology is particularly suitable for services that require centralized use of processors, such as application servers and communication servers. Many applications do not need to be rewritten after upgrading to the SMP platform.  

　○ SMPTechnology is the development direction of PC server in the future.

● Chassis technology

　　○ Vertical chassis
　　　　The vertical chassis is a computer cabinet with a height larger than the width (also known as a side vertical computer). The advantage of a vertical case over a horizontal computer is that it covers less floor space (desktop space). The height of the vertical chassis is usually between 18 and 27 inches. The miniature vertical chassis is about 14 inches.High and medium vertical chassis is usually about 16 inches.  

　　○ Base type
　　　　The pedestal chassis is usually wider and higher than the vertical chassis. Compared with the vertical chassis, the base chassis provides more flexible configuration options and scalability, and can generally provide heating plug-in and disk array functions.  

　　○ Rack mounting
　　　　The rack mounting system allows users to install multiple nodes or chassis on a metal rack and use the track to slide back and forth. The typical rack is 77 inches tall, 24 inches wide and 40 inches deep. The rack is measured in a vertical unit of measurement, expressed in the letter “U”.  

　　　　 l U＝l．75Inches or 4.445 centimeters. The 77 inch vertical rack is 40U.  

　　　　 The rack system consists of a display, a disk drive, an uninterruptible power supply (UPS), a network component, and a server node. The extended concept of the chassis is to divide the server nodes into separate parts that communicate over a server domain network, possibly with separate processing nodes and memory nodes., expand the bus node and disk array.

● Memory technology

　　The family of memory is also huge, with many different categories. According to the function of storing information, memory can be divided into RAM (Random Access Memory) and ROM (Read Only Memory, Read Only Memory). ROM isNon-volatile components, high reliability, stored in the ROM data can be permanent storage, without power off, so ROM is generally used to store system programs without modification or frequent modification, such as the motherboard BIOS program. According to the modifiable nature of information, ROM can also be divided into two parts.For MASK ROM, PROM, Flash Memory, etc., among them, MASK ROM, PROM belongs to the early products, the ROM family through a series of evolution, from the use of only once written ROM, the use of ultraviolet ray erasure of EPROM, the use of electricalThe EEPROM, which is cleared in this way, can be cleared by the common voltage on the motherboard. Now computers are developing very fast, and motherboard vendors need to upgrade BIOS frequently, so using Flash Memory to store BIOS programsAs the first choice, RAM is not only what we usually call memory, but also the main aspect we need to pay attention to. Here’s an introduction.  

　　○ RAMClassification
　　　　RAMIt is mainly used to store various field input and output data, intermediate calculation results, and exchange information with external memory and stack. Its storage unit can be read out or written or rewritten according to specific needs. Since RAM is made up of electronic components, it can only be used for temporary storage of programs and data.When the power is switched off or the power is cut off, the data will be lost, so it belongs to volatile components. Now RAM is mostly MOS type semiconductor circuit, it is divided into two kinds of dynamic and static. Dynamic RAM (DRAM) relies on the gate capacitance in the MOS circuit to memorize information. Because the charge on the capacitor will leak.Leaks need to be replenished regularly, so dynamic RAM requires Refresh, which takes extra time; static RAM (SRAM) uses bistable flip-flops to memorize information and saves data without repetitive refresh actions, so access is required.It’s much faster than DRAM. But dynamic RAM has higher integration and lower power consumption than static RAM, so it has lower cost and is suitable for large capacity memory. So Cache uses SRAM, and DRAM is usually used in main memory. The memory bar we usually touch isIt is made up of DRAM chips.  

　　○ DRAMSpecies
　　　　FPM DRAM（Fast Page Mode DRAM），That is, fast page mode DRAM. It is an improved DRAM, usually 30 line or 72 line (SIMM) memory. Basically, if the data you want to access in the system happens to be in the same column address or the same page (Page), the memory controller will not send it repeatedlyThe address is listed, and only the next row address is specified.  

　　　　EDO DRAM（Extended Data Out DRAM），That is to extend the data output DRAM. The speed is faster than FPM DRAM 15%~30%. It has the same architecture and operation as FPM DRAM, but shortens the waiting time between two data transfer cycles, so that the data in this cycle can be transmitted in the next cycle before it is completed.Speed up the processing of CPU data. EDO DRAM is currently widely used in computer motherboards, almost completely replacing FPM DRAM, working voltage is generally 5V, interface mode is 72 lines (SIMM), there are 168 lines (DIMM).  

　　　　BEDO DRAM（Burst EDO DRAM），EDO DRAM. It is an improved EDO DRAM. The difference between EDO DRAM and EDO DRAM is that EDO DRAM transmits only one set of data at a time, while BEDO DRAM operates in a “burst” mode and can transmit “batch” data at a time, generally BEDO DRAM can improve the performance of EDO DRAM by about 40%. Because of the emergence and prevalence of SDRAM, the social demand of BEDO DRAM has been reduced.  

　　　　SDRAM（Synchronous DRAM）That is synchronous DRAM. A kind of memory that is very popular at present. The working voltage is generally 3.3V, and its interface is mostly DIMM type of 168 lines. Its biggest feature is that it can be synchronized with the CPU’s external clock, and our CPU, motherboard use the same clock, if the CPU’s external clockThe working clock is 100MHZ, and the frequency sent to memory is 100MHZ. This will remove the time delay and increase the memory access efficiency.  

　　○ REGISTERED Memory
　　　Register IC The smaller IC chips (2-3 pieces) at the bottom of the memory bar play an important role in improving the driving capability. Server products need to support large amounts of memory, which cannot be driven by the current of the motherboard signal line alone. Instead, they use memory bars with Register, which pass through Register.IC improves drive capability, enabling the server to support up to 32GB of memory.  

　　○ ECCMemory
　　　　Error checking and correction memory (ECC) provides a powerful data correction system. ECC memory can detect not only one bit error, but also locate the error and correct the error before it is transmitted to the CPU, and transmit the correct data to the CPU. Allow the system to work uninterrupted and normal, ECCMemory can detect multiple dislocations (which parity memory can’t) and can generate alarms when multiple dislocations are detected, but it can’t correct multiple dislocations at the same time.  
　　ECCThe working process is as follows: when the data is written into memory, the ECC adds an additional bit of the data with an identification code. When the data is written back, the stored code is compared with the original code. If the code is inconsistent, the data is marked as “bad code”, and then the bad code is corrected and transmitted to the CPU.When multiple faults are detected, the system sends out alarm information.  

2. Types and functions of servers

(1) WWWServer (WWW Server)

　　 WWWServer, also known as Web Server or HTTP Server, is one of the most common and frequently used servers on the Internet. WWW servers provide users with web browsing, forum access, and so on.. For example: we are using browser access.
http://www.discuz.NETIn fact, you are visiting the WWW server of Discuz! To get the necessary forum information and web pages from the WWW server.  

（2）FTPThe server

   FTPServers are servers that provide users with various files (Files). FTP servers often store a large number of files, such as software, MP3, movies, programs and so on. Users can download the required text from the FTP server by logging in to the FTP server using FTP client software.Pieces and resources go to your own computer.
You can also upload files from your phone to FTP for other users to download to share file resources.  

（3）.Mail server

  e-mailIt is one of the most frequently used services on the Internet, and hundreds of billions of e-mails are sent and received every day on the Internet through mail servers. Like a post office, a mail server can provide users with receiving, storing and sending e-mail.  

   In addition to the three main servers described above, there are many other types of network servers, such as database server, proxy server, and Domain Name Serv.Er) wait…

 3.Server operating system

  At present, there are two main types of operating systems used in servers: Windows and Unix.

 (1) Windows   

　　WindowsIt is an operating system developed by Microsoft. In the server field, there are mainly Windows 2000 Server/Advanced Server/Data Center and Windows 2003 Standa.Rd Edition/EnterpriseEdition operating system, Windows has the advantage of operation.
Single, because Windows uses a graphical interface for operation, so the configuration of various server software functions is simple. But its shortcomings can not be ignored, for example, Windows exercises.
The cost of the system is relatively high, the security is relatively low, and the amount of traffic that can be taken is low.  

(2) Unix 

   UnixIt has a long history, and its categories and branches are complex. At present, the most widely used Unix system is Linux. Linux is not distributed by any company. Linux is developed and maintained by love programs and netizens from all corners of the world. Linux is completely free, and WinThe cost of Linux is 0 compared to dows.  
LinuxIn addition to the cost advantages, there are also many very good features, such as: extremely high performance, good stability, security and so on. At present, most large and medium-sized
Enterprises (including telecommunications companies and Google, Baidu, Sina, Sohu and so on) run servers on Unix / Linux systems.  

4. ApacheAnd IIS

 ApacheBoth IIS and WWW belong to the WWW server, which are the two most widely used servers in the world.  
(1) IIS 
 IISThe full name: Internet Information Server, developed by Microsoft, is part of the Windows operating system. IIS is a Web server that allows information to be published on Internet. IIS passThe hypertext transfer protocol (HTTP) is used to transmit information. IIS can also be configured to provide file transfer.
Protocol (FTP) service. FTP services allow users to transfer files from Web nodes or to Web nodes.

   IISThe features are simple configuration, friendly configuration interface, strong function, and provide support for ASP / ASP. net. But IIS’s performance and security are relatively poor, and IIS can only be used in Windows, not running in UNIX.  

(2) Apache 

   ApacheIt is the world’s No. 1 WWW server, and according to a survey conducted by Netcraft (www.netcraft.com), more than 60 percent of the world’s Web servers are using Apache.  

   1995The earliest Apache (0.6.2) was released in April 2004 by Apache Group, a non-profit entity that operates entirely on the Internet to determine the Apache Web serverWhat should be included in the standard release?  

Apache Characteristics:
1) It can run on almost all computer platforms (including Windows).
2) Powerful functional configuration;
3) Support general gateway interface (CGI);
4) Support for virtual host;
5) Support HTTP certification;
6) The proxy server is integrated internally.
7) Tracking ability of user session process;
8) Support FASTCGI;
9) Support Java SERVLETS;

● Common operating system

　○ MicrosoftWindows NT Server 4.0 Chinese / English
　○ MicrosoftWindows 2000/2003 Chinese / English
　○ SCO OpenServer5.0.5 
　○ SCO UnixWare7.1.1 
　○ Red Hat 6.2/7.0 
　○ TurboLinuxServer 6.1 
　○ SUN Solaris 7/8Chinese / English

○ Windows NT / Windows 2K/2003 
　· Integration with Windows client is better.
　· Provides certain documentation and application server compatibility capabilities.
　· Simplify installation and management, easy operation of the operating system and good user interface.
　· Provide more development tools, support from third party vendors is more.
　· At present, the growth rate of small and medium users is faster.
　· Directory management is difficult in large environments.
　· Compared with other operating systems, the reliability is poor.
　· After the configuration is changed, the system needs to be restarted.

○ SCO UNIX 
　· Scalability is better in high-performance RISC machines.
　· Easy to change network configuration
　· High safety and reliability
　· Providing built-in multi-user capability
　· The earliest and most widely supported Internet standard
　· The platform is extremely rich in application.
　· In the domestic financial and other important industries, there are more users.
　· Poor user interface, maintenance, management and use complexity.
　· No reliable development tools.

○ NetWare 
　· The performance of single CPU file servers is excellent.
　· High performance directory services enable easy management of large environments
　· More users are used in the early stage of the country.
　· Critical services have nothing to do with SMP.
　· Lack of support from third party vendors
　· No reliable development tools.

○ LINUX 
　· Free multitasking and multi-user operating system
　· Stable performance and small space occupation
　· It can run on Intel, SPARC and Alpha platforms.
　· No special technical support department.
　· The driving power of some devices is not perfect.

○ Solaris 
　· Various installation methods and high automation.
　· The ability to process data is very high.
　· It can interoperate with all kinds of platforms.
　· Software is expensive.
　· Weak support for Intel based server technology

 

This article is reproduced from http://blog.csdn.net/dragonpeng2008/article/details/6642865