Five major server procurement problems need to be considered in advance

Five major server procurement problems need to be considered in advance-Linux Enterprise Application-Linux server application information. The following is a detailed description. There are a wide variety of products in the server market, and various manufacturers have begun to provide services and solutions, making users more and more confused when purchasing Server products. The answers to the following questions are believed to provide some reference when users buy servers, because spring creates the right server product in the autumn, it is a necessary factor to ensure normal business operation.

Question 1: What are the main configuration parameters when purchasing an IA server?

The type, frequency, and quantity of CPU and memory CPU determine the performance of the server to a considerable extent. The server should adopt dedicated ECC memory verification and use it with different CPUs.

Even if the same chipset is used for the chipset and motherboard, different motherboard designs will have an important impact on server performance.

The NIC server should be connected to the port with the highest transmission rate and have at least one gigabit Nic configured. For some servers with special applications (such as FTP, file server, or VOD Server), you should also configure two Gigabit NICs.

The read/write rate of the hard disk and RAID card determines the processing speed and response speed of the server. In addition to using IDE hard disks on entry-level servers, SCSI hard disks with higher transmission rates and better scalability are usually used. For some servers that cannot be easily terminated, hot swapping of hard disks should also be used to ensure that the servers are maintained and resized without downtime.

Disk redundancy uses two or more hard disks to implement disk arrays. Nic, power supply, fan, and other components redundancy can ensure that the server can still run normally after some hardware is damaged.

Hot swapping refers to live insertion and removal of hard disks or boards for fault recovery and system resizing.

At the same time, when selecting an IA server, you often need to consider performance indicators such as manageability, availability, scalability, security, and reliability.

Question 2: What is the application scope of a 64-bit server?

Here we will talk about some new 64-bit servers, such as anteng and AMD64. From the perspective of application types, it can be roughly divided into the main Domain Server, database server, Web server, FTP server and email server, and high-performance computing cluster system.

Master Domain Controller: The Management Center of networks, users, and computers, providing a secure network environment. The System Bottleneck of the master domain controller is the memory, network, CPU, and memory configuration.

File Server: a file server is a network data storage warehouse. Its performance requirement is to quickly transfer data between users on the network and the server disk subsystem.

Database Server: The Database Engine includes DB2, SQLServer, Oracle, and Sybase. The Database Server generally needs a multi-processor system. Taking SQLServer as an example, SQLServer can take full advantage of SMP technology to execute multi-threaded tasks. By using multiple CPUs, can perform parallel operations on the database to increase the throughput. In addition, SQLServer has 90% clicks on the L2 cache, so the larger the L2 cache, the better. Memory and disk subsystems are also crucial for database servers.

Web server: the Web server is used to respond to Web requests. Its performance is determined by the website content. If the Web site is static, the system bottlenecks are network, memory, and CPU. If the Web server primarily performs intensive computing (such as dynamically generating Web pages), the system bottlenecks are: memory, CPU, disk, and network, because these websites use the dynamic content connected to the database to generate transactions and queries, this requires additional CPU resources, you must have enough memory to cache and process dynamic pages.

High-performance computing cluster systems: generally four or more nodes, the node machines use Opteron Systems Based on the anteng and AMD64 technologies, the performance of this cluster system depends on the technical strength of the manufacturer, the design of the cluster system, and the application optimization.

Question 3: What are the policies for purchasing multi-processor servers?

Before purchasing a multi-processor system, you must understand the workload, select the appropriate application software and operating system, and then determine the servers that enable them to run. It is worth noting that you 'd better purchase a server with a higher computing power than you currently need to adapt to future expansion needs.

First, the selection of the processor is closely related to the selection of the main operating system platform and software. You can select a type of processor, such as Linux and Linux, which are applied to operating systems such as SunSolaris, IBMAIX, and Linux. Most users also use Intel processors for price and operating system considerations.

Secondly, select an appropriate I/O architecture. Currently, the most common bus structures are PCI and PCI-X. PCI is evolving rapidly to include 32-bit and 64-bit data channels, and provides support for 33 MHz and 66 MHz clock speeds. PCIExpress is a new serial technology, which completely revolutionizes the original parallel PCI technology and is compatible with PCI technology. The PCIExpress bus uses point-to-point technology to allocate dedicated channel bandwidth to each device without sharing resources between devices. Fully protect the bandwidth resources of each device.

Then, select the appropriate memory. Most multi-processor systems currently support correction of SDRAM.

Finally, it is the storage problem. The number of drive tanks supported by the server will inevitably affect the shape and height of the server. If the server is connected to the SAN, there are not many requirements for internal storage. However, if the device is located remotely without a SAN, you can purchase a system that supports up to eight externally accessible hot swapping SCSI drives.

Question 4: What is the application of blade servers?

The blade server was initially positioned as a service provider and a large enterprise looking to compress a large amount of computing power into a small space. Currently, many system vendors sell these slice servers by integrating data center infrastructure, removing miscellaneous cables, optimizing management, and cost-effective. The size of the blade server is only a fraction of that of the standard 1U server and requires less power. It is installed in a dedicated machine box that allows them to share resources.

Question 5: will the cost of the blade server be reduced in addition to the return on computing density?

Experts believe that deploying the blade server will be rewarded with space savings. When using a blade server, the computing capacity of each rack reaches 10 GHz. When using a traditional platform, the actual computing capacity of each rack is GHz, this is a 20-fold improvement. The cost of data center space is very expensive, and this is the place where the use of blade servers has a huge return: computing density.

However, early adopters also pointed out that blade servers are not applicable to everyone. Some manufacturers say that you must have blade servers. They will replace all servers with blade servers. Users should use it in the most appropriate place. If you try to use space more efficiently, you should consider choosing a blade server.