In the business and enterprise network, generally are mining C mode, that is, "client-server" mode, according to the authority of the investigation and prediction, the future of the network development also shows the trend of centralized computing. And the rise of data center to the server configuration density put forward more and more high requirements, it is in this situation, in order to break the limit of individual performance of the server, a new generation of clustered server-blade server.
With the further development of large scale computing and network server's demand for parallel processing capability, the blade server with "collective power" casting powerful processing ability has been valued more and more. Blade server is a major breakthrough in the development of server technology in recent years. Its significance is much greater than the increase in processor frequency and memory, hard disk and other storage media capacity, why blade server has such an important position, the following author on the basis of their own understanding, combined with some relevant technical information for you to explain:
First of all, in the business and enterprise networks, generally are mining C mode, that is, "client-server" model, according to the authority of the investigation and prediction, the future of the network development also showed a trend of centralized computing. And the rise of data center to the server configuration density put forward more and more high requirements, it is in this situation, in order to break the limit of individual performance of the server, a new generation of clustered server-blade server.
Why is it called Blade Server, this is based on the characteristics of the product profile named, because the blade server components are flat and flat shape, like a blade, so the product launch when the manufacturers are also naturally named it; We usually say that the blade server is in the standard height of the rack-type chassis can be inserted in multiple card server unit (that is, blades, it is actually an industry-standard board with processors, memory, and hard drives, and the operating system is installed, so a "blade" is a small server. , this piece of blade combination, data interchange and sharing, in the coordination of system software synchronization work can become a high availability and high-density new servers.
Blade servers are widely used, especially for computational-intensive applications such as weather modeling, data acquisition, simulation, digital image design, aerodynamics modeling, and so on. And for industry applications, such as telecommunications, finance, IDC/ASP/ISP applications, mobile phone base station, video-on-demand, web host operations, laboratory systems, etc., blade Server can also be able to work. At present, the main application in the following two areas:
1, as a front-end Web server
This approach is easier to understand and can give full play to the advantages of high server density, clustering, and remote management. Content service providers can use blades to form high-density clusters that enable high volume Web servers to be connected to high-end servers or clustered systems as database servers. Storage service providers can use the same front-end scenario, with the back-end matching NAS devices to provide storage services. The advantage of blade server in this kind of application is that it occupies less space and can save the trust cost effectively compared with ordinary rack server.
2, as a small and medium-sized network server
This way can give full play to the blade server easy to manage, flexible configuration and scalability of good advantages. An SME network that uses Ethernet to connect to the Internet. There are approximately 300 client nodes, the desktop uses Windows 98 systems, and the network requires several servers, including file servers, CD-ROM servers, proxy servers, intranet servers, DHCP servers, domain servers, DNS servers, mail servers, web/ftp servers, The database server and the hub print server. Some of these services can share servers, some may need to be clustered to meet performance and availability requirements, with 2 12-port hundred-gigabit switches, a single rack NAS device, and a blade server to meet these requirements, of which 2 switches are intranet switches and extranet switches respectively. For enterprises, this high-density not only saves valuable cabinet space, but also saves the wiring cost, and can save electricity, thus reducing the demand for UPS.
In general, the blade server has the following characteristics, and is almost the advantage of it:
1. High density calculation
High-density is the biggest feature of the blade server, regardless of whether it is high or low in the field of work, but only in terms of computational density. Of course, computing power is not only determined by the amount of processor stack, but the powerful computing power of a single system is not reduced by 4~8 processors alone. So what does the blade rely on to occupy the market? Overall, whether it's the operating system or the application software, are significantly behind the pace of hardware development, distributed applications such as the Web are closer to the distribution of computing--a large number of processes in parallel at the same time, and a single process with very low computational processing requirements, which is consistent with the blade's emphasis on high-density decentralized computing architecture. For blades, the increased computational density allows data centers to serve more customers with less space, and to reduce hosting costs for Web sites.
2. Distributed computing
Even compared with the group-level server, blade because of the configuration of a single blade is downwind, so the need for load balancing/clustering software to achieve good results, the applicable domain is also limited to distributed high-density computing environment: IDC, WEB, domain name resolution.
3. Low power consumption
Tower Server because of the most abundant internal space, can be used to emphasize a single performance of equipment, regardless of its volume and power consumption, as long as can be in sufficient space to solve the problem of heat dissipation can easily load. Rack-type server not only in the shape of the reduction, but more importantly, they are standard stature, can be stacked in a rack to install, significantly improve the space utilization, improve the unit volume within the computational density. However, it is still based on traditional server components, with no reduction in power consumption and fever, but only through more optimized thermal design to meet job needs.
The blades overcome the above drawbacks, taking RLX System 324 as an example: a 7-foot-high rack, which can be put into 40 1RU (1.75-inch) servers, can be put into 336 rlx blades. This incredible density is due to the integration of the Transmeta Crusoe (CRUSOE) processor. The benefits of using Crewe chips are low power consumption and power saving. The average power consumption of each piece of the Crewe Blade is only about 7 watts at the time of execution. (The maximum power supply for each RLX blade is 15 watts, while the traditional server is 75 to 120 watts.) The RLX approach not only saves you a lot of electricity, but also saves you expensive hardware purchases, such as no power off system (UPS), generator systems, and cooling devices.
4. Low cost
The blade server can be considered as the hardware of all server systems: Processor, memory, hard drive, and network connection device one by one to a single expansion card. Such consolidation represents a more attractive alternative: it lowers costs, lowers temperatures, and reduces the amount of space needed, further improving server stability and lower overall cost of ownership (TCO).