Server Basics Answer

1 Dual

Question: I often hear about the two-way Xeon XX server. Recently, there have been dual-core Xeon, both CPUs. Is it dual-core equal to dual-core?

Answer: No

Regardless of the single, dual, four or even eight way of the server, the "road" refers to the number of physical CPUs of the server, that is, the CPU on the server board. The number of slots.

The recent emergence of dual-core processors is to encapsulate two CPU cores inside a physical CPU. This has the advantage of allowing users to have more powerful performance without increasing costs. Moreover, it can significantly reduce the performance-to-power ratio, which is also of positive significance for enterprise users to save on the cost of use.

2 The difference between Xeon and Pentium

Question: In many servers, there are strong processors, and Pentium 4 as a processor, except for Pentium 4 Processor server products are cheaper, what is the difference between Xeon and Pentium?

Intel Pentium 4 631 3.0GHz (box)

Intel Xeon 3.2G (800MHz/2M/box)

Answer: The Xeon processor used on the server is basically the same as the normal P4 processor.

One of the differences is that the Intel Pentium 4 processor began to classify the Pentium 4 processor as a personal processor. Users can't build a 2-way server system with two Pentium 4 processors to develop computing power. Higher Xeon processor. Xeon processors are currently classified into Xeon DP (maximum support for dual processing) and Xeon MP (maximum support for 8-way processing).

The second difference is that the secondary cache is different. Xeon's L2 cache is 1MB to 16MB, P4's L2 cache is 512KB to 1MB, and the L2 cache capacity is also an important factor in determining server processing performance.

The Xeon series CPUs are mostly 604 interfaces, while the P4 CPUs are mostly 478 pins or 775 architectures.

Note: In the X86 architecture, in addition to the Intel Pentium and Xeon series processors, AMD also provides a dedicated processor for the server - Opteron Opteron, which can support up to 8 channels.

AMD Opteron 865 865 (散)

3 types of servers

Question: The server is divided according to structure, there are several types?

Answer: At present, according to the shape structure, the server can be divided into three types: tower type, rack type and blade type.

Tower Server

The general tower server chassis is similar to our usual PC chassis, while the large tower chassis is much larger. In general, there is no fixed standard for the form factor.

Rack Server

Rack Server is a server type with a 19-inch rack as a standard width due to the dense deployment of the enterprise, and the height is from 1U to several U. Placing the server in a rack is not only good for routine maintenance and management, but it can also avoid unexpected failures. First, placing the server does not take up too much space. The rack server is neatly discharged into the rack without wasting space. Secondly, the connecting wires and the like can also be neatly placed in the rack. Power cables and LAN cables can be routed in the cabinet to reduce the number of cables that are stacked on the ground, thus preventing accidents such as kicking off wires.

The specified size is the width of the server (48.26cm = 19 inches) and the height (multiple of 4.445cm). Since the width is 19 inches, the rack that satisfies this requirement is sometimes referred to as a "19-inch rack."

Blade Servers

Blade Servers is a low-cost HAHD (High Availability High Density) platform designed for specific application industries and high-density computer environments. Each of these "blades" is actually a system motherboard, similar to a separate server. In this mode, each motherboard runs its own system, serving a different set of users, and is not related to each other. However, you can use system software to aggregate these boards into a single server cluster. In cluster mode, all motherboards can be connected to provide a high-speed network environment that can share resources and serve the same user base.

4 What is "U"?

Question: What do you mean by "U" in 1U and 2U servers? Is it the number of server processors?

Answer: “U” refers specifically to the thickness of the rack server in the server field. It is a unit that represents the external dimensions of the server and is an abbreviation of unit. The detailed size is determined by the American Electronic Industry Association (EIA) as an industry group.

The reason for specifying the size of the server is to keep the server in the proper size for placement on an iron or aluminum rack. The rack has a screw hole for fixing the server, and it is aligned with the screw hole of the server and fixed by screws.

The thickness is 4.445cm as the basic unit. 1U is 4.445cm, and 2U is 2 times that of 1U is 8.89cm. In other words, the so-called "1U rack server" is a product that meets the EIA specifications and has a thickness of 4.445 cm.

5 Differences between servers and workstations

Question: What is the difference between servers and workstations?

Servers provide various services to workstations, network communication services, file sharing Services, hardware sharing services, various resource services. The workstation can also help the server to offload computing and other tasks while acquiring various resources of the server.

Servers and workstations are high-performance computers, but the server is relatively focused on data throughput, so more peripherals (hard disks, I/O slots, etc.) are supported, while workstations focus on Graphics processing capabilities, so the peripherals are relatively small, but using a framework designed specifically for graphics processing, using high-end graphics cards, support for 3D image processing. Workstations are mainly used in various design and multimedia production areas.

Central processing unit, CPU is the heart of the computer, including computing components and control components, is the core to complete various operations and control, and is the most determined computer performance. Important parts. The main parameters are the main frequency of the work and the number of bits of data transferred or processed at one time.

Hard Disk:

Hard disk is the most important place to store data. Most of the information needed for execution is mostly read from the hard disk. Once the hard drive is damaged, you can't use any information, so the hard drive is the most important storage device. The larger the capacity of the hard disk, of course, the more data the server can store.

Types of Hard Disks: Since IBM produced the first hard disk, the mainstream hard disks currently used in servers are divided into two types according to the interface: SCSI interface hard disk for medium and high-end storage and entry-level desktop computer. SATA interface.

Hard disk capacity: I often hear hard disk space is 80GB, 160GB (usually SATA interface); 73GB, 146GB (SCSI hard disk capacity). GB is the unit of capacity, 1GB is equal to one million bytes (Byte), and is also equal to 1000MB. The larger the capacity of the hard disk, the better. The larger the capacity, the more data can be stored.

Hard disk speed: Hard disk speed refers to the speed of the hard disk motor. The faster the speed, the faster the speed of reading and writing data. Therefore, the faster the speed of the hard disk, the price will be higher. The current hard disk speeds are common to 5400 rpm and 7200 rpm.

Memory:

DDR: DDR (Double Data Rate) SDRAM provides dual memory bandwidth and is more efficient than SDR (Single Data Rate) SDRAM. Currently, mainstream memory is already faster DDR2 memory.

RAM: Random Access Memory (RAM): RAM is a memory that can be read and written. When we write data to RAM memory, we can also read data from RAM. ROM memory is different. However, RAM must maintain its stability by stable and smooth power, so once the power is turned off, the original data in RAM will disappear.

Shorthand for ECC: Error Checking and Correcting. It is widely used in computer instructions in various fields to correct errors through memory, and the computer's operating instructions can continue to execute.

The most common mistakes in memory are: unit error, multiple dislocations, column errors, and errors. They are all similar. Most unit errors occur when there is a bit error when reading a complete bit or word. When reading the same bits and words, always the same bit of data is wrong, it is called multi-dislocation. Unit errors occur in many words, and they are called wrong or wrong.

How these errors are corrected: ECC memory uses extra bits to store a code encrypted with data. When the data is written to memory, the corresponding ECC code is also saved. When the data just stored is read back, the saved ECC code is compared with the ECC code generated when the data is read. If the two codes are not the same, they are decoded to determine which one in the data is incorrect. This error bit is then discarded and the memory controller will release the correct data. The corrected data is rarely put back into memory. If the same erroneous data is read again, the correction process is executed again. Rewriting data increases the overhead of the process, which can result in a significant reduction in system performance. If it is a random event rather than an error caused by memory, the error data of this memory address will be replaced by other data written again.

RAID:

RAID (Redundant Array of Independent Disks) storage technology can improve the performance and reliability of data storage subsystems. The original purpose of developing RAID was to distribute large amounts of data across small disk arrays, which seemed to be a "logical" hard disk for the server's operating system.

Intel's integrated RAID controllers and software help RAID applications in e-commerce applications in storage subsystems (including 5, 8 or more independent disks). Choosing RAID means that even if a hard drive fails, the server and network remain fully operational without losing data. In the early stages of RAID development, the "I" in RAID was "price economy." With the sharp decline in the cost of large-capacity hard drives, the industry gradually changed the meaning of the letter "I." It now means "stand-alone." "This illustrates one of the main advantages of the RAID subsystem: stand-alone disks, the ability to protect data, improve network storage performance, and keep the server up and running.

A hard drive failure can cause server downtime, network Performance degradation and disruption of e-commerce applications, properly configured RAID arrays help prevent such things from happening.

Selecting RAID means that even if the hard drive fails, the server and network remain fully operational. There is no data loss. In some cases, users on the network will not even notice that the hard disk has failed.

With the "hot plug" function supported by Intel integrated RAID software, the server can be normal. Disassemble and replace failed hard drives at runtime. The continued decline in the price of fast, high-capacity hard drives makes RAID a more storage technology for e-commerce businesses.

Traffic:

Network traffic reflects the number of packets passing through the network and the degree of congestion.

Byte: is a unit of measurement file size, which is usually used to describe the drive or The size of the disk, 8 bits = 1 byte. //This article is transferred from www.45it.com Computer Software and Hardware Application Network

Bandwidth:

Used to measure the connection of a server to the Internet. Data transmission speed in bps (bit per second)

Input bandwidth: From the server's point of view, we define the Inbound Bandwidth as the speed at which data can be transmitted from the Internet to the server.

Output bandwidth: From the server's point of view, we define the output bandwidth (Outbound Bandwidth) as the speed of data transfer from the server to the Internet.

The bandwidth measurement unit is: Bit, Bit is called bit, It is the smallest unit of computer memory, 1Bit has 0 and 1 changes.

What is the basic architecture of SAN, NAS? SAN?

The SAN is almost independent of the server network system. Unlimited storage Fast storage network, which uses Fibre Channel as the transmission medium and Fibre Channel and SCSI application protocol as the storage access protocol to network the storage subsystem.

Fibre Channel technology has high bandwidth and low bit error rate. The long distance feature is especially suitable for mass data transmission, so it is applied to the connection channel and networking technology between the host and the memory.

NAS: Network Attachment Storage (NAS)

This technology can meet the host storage needs of a dedicated direct attached storage device. Because NAS has the characteristics of protocol disclosure, simple operation and wide adaptability, especially in the multi-user network computing environment based on file processing, NAS has become an important storage means with its good expansion capability.