Top 10 servers

1. Cluster Technology

Cluster technology is a new high-performance computing technology in recent years. It is a single computer system composed of a group of independent computers through high-speed communication networks and managed in a single system mode. Its starting point is to provide high reliability, scalability, and disaster resistance.

A server cluster contains multiple servers with shared data storage space, and each Server communicates with each other through an internal lan. When one of the servers fails, the applications it runsProgramIn most cases, all computers in the cluster share a common name, and any server in the cluster system can be used by all network users.

The servers running in the cluster system are not necessarily high-end products, but the server cluster can provide high-performance, non-stop services. Each server can undertake some computing tasks, in addition, because the performance of multiple servers is clustered, the overall system computing capability will be improved. At the same time, each server can undertake a certain degree of fault tolerance tasks, when a server fails, the system can isolate the server from the system with the support of specialized software, and achieve new load balancing through the load transfer mechanism between servers, sends an alarm to the system administrator.

The cluster system achieves high availability and high reliability through function integration and Failover technologies. The cluster technology also provides relatively low total cost of ownership and powerful and flexible system expansion capabilities.

2. SMP (Multi-processing) Technology

Symmetric Multi-Processing refers to a group of processors (multiple CPUs) on a computer ). The memory subsystem and bus structure are shared among CPUs. Although multiple CPUs are used at the same time, from the management point of view, they behave like a single machine. The system distributes task queues evenly across multiple CPUs, greatly improving the data processing capability of the entire system. With the improvement of the user's application level, it is indeed difficult to use a single processor to meet the needs of practical applications. Therefore, server vendors have adopted Symmetric Multi-processing systems to solve this conflict. The most common Symmetric Multi-processing system in PC servers usually uses 2, 4, 6, or 8 processors. At present, Unix servers support a maximum of 64 CPUs, such as Sun's enterprise 10000 product. The most critical technology in SMP systems is how to better resolve the issue of inter-communication and coordination between multiple processors.

3. NUMA (non-uniform memory access) distributed memory access

In the field of high-performance computing, numa is a new technology widely used by various manufacturers. The idea is to combine SMP and cluster advantages. It is a system composed of several independent nodes connected through a high-speed dedicated network. Each node can be a single CPU or an SMP system. This technology is an improvement to the traditional intel-based SMP system. Traditional intel-based SMP systems often block data because the data on the shared memory bus is too crowded. In general, they cannot accommodate 16 ~ 32 processors. If NUMA technology is used, each intel processor will have its own local memory and can form a static or dynamic connection with the memory in other chips. The NUMA server can accommodate 64 or more processors. The NUMA system structure's machines are distributed in memory as a whole from the internal perspective. However, because of its high transmission speed, users use it just like machines with shared memory. The price is between the SMP system and the cluster system. Initially, numa technology was built on a dedicated IRIX operating system and MIPS processor. Today, this technology has been adopted by more and more vendors.

4. ISC (Intel Server Control) Intel Server Control

ISC is a network monitoring technology that is only applicable to servers with integrated management boards using Intel architecture. With this technology, you can monitor all servers using Intel motherboard on the network on a common client, and monitor and judge whether the server is "healthy ". Once an error occurs in the server's chassis, power supply, fan, memory, processor, system information, temperature, voltage, or any of the third-party hardware, an alarm is reported to the Administrator. It is worth mentioning that the network between the monitoring terminal and the server can be either a LAN or a WAN, and the server can be started, closed, or reset through the network, this greatly facilitates management and maintenance.

5. EMP (Emergency Management port) Emergency Management Port

EMP is an interface on the server board for remote server management. The remote control machine can be connected to the server through MODEM, and the control software is installed on the control machine. The remote control machine can perform the following operations on the server through the EMP console control interface:

A. Turn on or off the server power.

B. Reset the server, and even include the bios and CMOS parameters of the motherboard.

C. Monitor internal server conditions, such as temperature, voltage, and fan conditions.

The above functions allow technical support personnel to promptly resolve many server hardware faults through MODEM and telephone lines. This is a good technical means to achieve fast service and save maintenance costs.

The ISC and EMP technologies can be used to remotely monitor and manage servers.

6. Raid (Redundant Array of Independent Disks) cheap redundant disk array

Because the disk access speed cannot keep up with the CPU processing speed, the disk becomes a bottleneck for improving the server I/O capability. In order to solve the increasing contradiction between the high speed of the computer CPU and the low speed of the disk, Professor David. A. pattorson of UC Berkeley proposed the concept of RAID in 1987. The technical idea is: using existing small and low-cost disks, multiple disks are combined into a disk array in a certain way, through some hardware technology and a series of schedulingAlgorithmThe entire disk array is like a large disk with a large capacity and high reliability and speed.

RAID technology uses several hard disk drives to form a whole according to certain requirements. The entire disk array is managed by the array controller. Disk arrays have many features: firstly, they increase the storage capacity; secondly, multiple disk drives can work in parallel to improve the data transmission rate; thirdly, the verification technology improves the reliability: if one hard disk in the array is damaged, the original data on the damaged disk can be restored using other disks without affecting the normal operation of the system, the damaged hard disk (that is, the hot swapping function) can be replaced in the live state. The array controller automatically writes the restructured data to the new disk, or write a hot backup disk to create a new hot backup disk. In addition, the disk array is usually equipped with redundant devices, such as power supplies and fans, to ensure the heat dissipation of the disk array and system reliability.

Commonly used raid types include raid0, raid1, raid3, and RAID5.

7. I2C (Inter-Integrated Circuit) Bus

I2C bus is a serial bus developed by Philips. It was originally developed for audio and video devices in 1980s and is mainly used in server management. The I2C bus includes an interface with two ends. Through an interface with a buffer, data can be sent or accepted by I2C. Control and status information are transmitted through a set of memory ing registers. The I2C hardware bus technology can be used to centrally manage all components of the server and monitor multiple parameters such as memory, hard disk, network, and system temperature at any time, increasing the system security, convenient management.

The main advantage of I2C bus is its simplicity and effectiveness. Because the interface is directly on the component, the space occupied by the I2C bus is very small, which reduces the space of the circuit board and the number of chip pins, thus reducing the interconnection cost. The bus can be up to 25 feet in length and support 40 components at a maximum transmission rate of 10 Kbps. Another advantage of the I2C bus is that it supports multi-mastering, and any device that can send and receive data can become the master bus. A main bus can control signal transmission and clock frequency. Naturally, there can be only one master bus at any time point.

8. Intelligent Input/output (Intelligent i2o) Technology

The I/O system of PC servers is designed for single-user PC desktops instead of dedicated servers for processing large throughput tasks. Once a network center device is built, the data transmission volume increases significantly, therefore, I/O data transmission often becomes the bottleneck of the entire system. I2o smart input/output technology assigns tasks to smart I/O systems. In these subsystems, the dedicated I/O processor will be responsible for tedious tasks such as interrupt processing, buffer access, and data transmission. This improves the system throughput, the server's main processor can also be freed to handle more important tasks. Therefore, PC servers implemented according to i2o specifications can handle more tasks without changing the hardware scale, as the core of Small and Medium-sized networks, low-end PC servers can improve performance.

9. multi-processor communication and coordination Technology

Enterprise-level PC servers are mostly multi-processor structures, so communication and coordination between multiple processors is very important. With multi-processor communication and coordination technology, PC servers can divide more than four processor groups into multiple groups, each with a high-speed cache system, by checking the consistency of the cache ing structure, the high-speed cache information built in each group of processors is consistent with the corresponding information in the memory during the computing process. To ensure high-speed communication between systems, high-speed switching module design is also widely adopted in the system using this technology, so that each group of processors in the system can exclusively occupy a MHz system bus. In some systems, multiple independent Memory Boards are also used. Each memory board occupies a single MHz system bus, A high-speed exchange bus system is used between these memory boards, multi-group processor modules, and I/O bus to ensure that any group of devices can communicate with each other at a high speed of MHz, in this way, the transmission bandwidth of the entire system reaches a high level. These measures not only effectively solve the bottleneck of transmission bandwidth in traditional multi-processor systems, but also greatly improve the overall performance of the system, it also provides a stable upgrade solution for the system cluster and a hardware platform with higher performance and better availability for enterprise key operations.

10. Hot Swap)

The hot swapping function allows you to remove and replace damaged hard drive, power supply, or board components without shutting down the system or powering off the power, this improves the system's timely disaster recovery capabilities, scalability, and flexibility. For example, some high-end application-oriented disk image systems can provide hot swapping of disks. If there is no hot swapping function, even if the disk is damaged, it will not cause data loss. You still need to temporarily disable the system to replace the hard disk, using Hot swapping technology, you can simply turn on the connection switch or turn the handle to directly retrieve the hard disk, and the system can still operate normally without interruption.

For some servers used in key tasks, the damaged RAID card or ethernet card can be replaced without stopping services, which greatly reduces the system downtime caused by hardware faults.