The so-called blade server refers to the standard height of the rack-mounted chassis can be inserted into a number of card-type server units, to achieve high availability and high density. Each "blade" is actually a system motherboard. They can start their own operating systems, such as Windows NT/2000, Linux, etc., through the "on-board" hard disk, similar to a separate server, in this mode, each motherboard runs its own system, serving a designated group of different users, there is no correlation between each other. However, administrators can use system software to assemble these motherboards into a server cluster. In cluster mode, all motherboards can be connected to provide a high-speed network environment and simultaneously *** enjoy resources to serve the same user groups. By inserting new
"blades" into the cluster, overall performance can be improved. And because each blade is hot-swappable, systems can be easily replaced and maintenance time minimized.
These blade servers are designed with low power consumption, small space, and low stand-alone price, and it also inherits some of the technical specifications of traditional servers, such as hot-swap and redundancy into the blade servers, which are designed to meet the demands of dense computing environments on server performance; some also have built-in load balancing technology to effectively improve server stability and core network performance. Some also have built-in load balancing technology to effectively improve server stability and core network performance. From the outside, compared with traditional rack/tower servers, blade servers can maximize server space and cost savings, and provide users with flexible and convenient means of expansion and upgrading.
Characteristics of Blade Servers
There are two recognized characteristics of blade servers: one is that they overcome the shortcomings of chip server clusters, and have been made the end of clusters; the other is that they enable cabinet optimization.
Cluster Terminator
As we all know, as a load balancing technology, server clustering has been widely adopted in effectively improving the stability of the system and the performance of the core network services. In a clustered system, if you want to provide a higher-end computing and service performance, you only need to add more units to get a higher performance. More importantly, server clustering also provides redundancy and fault tolerance for any individual server.
The IT industry is currently in the midst of a major effort to develop powerful and reliable computers that are adapted to broadband networks. Over the past few years, broadband technology has greatly enriched the transmission of the information superhighway. The birth of server clustering and RAID technology provides a new solution for Internet applications of computers and data pools at a much lower cost than traditional high-end dedicated servers and mainframes. However, server clusters have low integration capabilities, and managing such clusters causes many administrators significant headaches. Especially with the increasing demand for cluster expansion, the amount of work required to maintain these servers is simply unimaginable, including the internal connections between servers and the space requirements for placement. All of these physical factors limit cluster expansion. The advent of blade servers is a timely solution to these problems. In cluster mode, all motherboards of a blade server can be connected to provide a high-speed network environment, **** enjoy resources. At the same time, each blade can be built-in monitor and management tool software, configuration of a high-density server can solve the management of one to one hundred servers, if you need to add or delete servers in the cluster, just insert or pull out a board can be, to minimize the maintenance time. In this sense, Blade Server fundamentally overcomes the shortcomings of server clustering.
Enabling Cabinet Optimization
In one sense, Blade Server takes the natural leap to cabinet optimization. Blade servers increase the density of space occupied by cabinet-based servers by another 50 percent. Data shows that in the cabinet system configuration under the premise of 1U rack-optimized server system transplanted to the blade server, the space occupied is only the original 1 / 3 ~ 1/2. And in a standard cabinet environment, the processing density of the blade server to increase four to five times. For example, in the processing of 1024 nodes of high-density computing server environment, 1U configuration requires 24 cabinets, which does not include the Ethernet switching hub occupied by the cabinet space, while the use of inserted eight "blade" blade servers, only 9 cabinets are required, but includes the space of the Ethernet switch. In the same amount of square footage, a data center can achieve eight times the server leasing revenue by deploying blade servers as compared to rackmount servers.
Additionally, blade servers are centrally managed, which simplifies server administration. In a world where IT staff is increasingly scarce, organizations that adopt blade servers can reduce maintenance costs by hiring fewer highly paid server administration and maintenance staff. Also, the low-power design of blade servers significantly reduces energy consumption, saving energy while reducing costs.
As an emerging server product, readers may still lack an intuitive understanding of it. Each blade server typically consists of a cabinet and blades, so the blade server logo consists of the cabinet's model number and the blade's model number **** the same, unlike previous servers that were represented by a single server model. The blades are connected to the cabinet through the CompacPCI interface on the cabinet's back panel. Server cabinets can typically hold from eight to dozens of blades. Blades are dominated by server blades, and each server blade is a fully functional server.
Here, we introduce you to a common kind of blade server to understand its basic composition.
Depending on the server functions to be performed, blade servers are divided into server blades, network blades, storage blades, management blades, Fibre Channel SAN blades, extended I/O blades, and so on for different functions of the corresponding blade servers.
Currently the most common server blades generally use 1 for the Intel Pentium Ⅲ processor, and the use of ServerWorks LC-E chipset, Intel 815 chipset, Via Pro266 chipset, support for the memory capacity and type of the chipset to determine the type of memory is generally with the ECC function of the SDRAM or DDR. DDR. Due to the more serious heat dissipation problems of blade servers, some vendors have also adopted the low-power Transmeta 5600 processor in their designs. Currently, HP, Sun is also working to make their RISC processors into server blades, just not yet available.
In addition to connecting the cabinet backplane interface, the server blade generally also has a PMC expansion interface, you can connect the PMC interface expansion card, such as SCSI cards, fiber-optic memory cards, etc., its function is equivalent to the PCI expansion slot, just the corresponding interface expansion card price is slightly expensive. The server blade uses a 65mm (2.5-inch) hard disk with the same specifications as the laptop, and generally only installs the operating system and simple application software with lower performance.
Network Blades
Network blades are the functional equivalent of LAN switches, providing good network monitoring and management. Network blades generally provide 10/100Mbps ports to connect server blades with twisted-pair cables to provide high-speed uplinks (Gigabit ports) to the outside world. Blade servers with NAS storage are often equipped with two network blades, one of which is dedicated to connecting to the NAS device. Each blade supports 10/100/1000M Ethernet connectivity and can be backplane-mounted with a 10/100/1000M Layer 2-4 switch, which connects the blades mounted in each slot in the system to the switch, providing an IP-based switched network. By integrating this bus, the blade server system can be well integrated with IP services and voice services to provide a variety of different value-added telecom services.
Storage Blades
A storage blade can be thought of as a hard disk module that provides storage functionality to the server blade via a backplane bus or hard disk interface cable. Storage blades are typically equipped with two high-performance 90mm (3.5-inch) hard drives with IDE, SCSI, and Fiber Channel interfaces.
Management Blades
The KVM (Keyboard, VGA, Mouse) blade of the first generation of blade servers is arguably the simplest of the management blades, providing management control of all server blades.The KVM blade, which provides the keyboard, mouse, and monitor interfaces, and the KVM blade often includes a floppy drive and CD-ROM drive, making it easy for users to operate the server blade directly. Server blades. KVM blades provide toggle switches for switching between blades on a cabinet or between cabinets. Second-generation blade servers have more powerful management features, but they vary from product to product. The management blade often provides centralized monitoring and management of one or more blade servers through a monitoring and management chip integrated on the server blade. The management blade provides the necessary configuration information to the other blades in the server cabinet, and receives alarms and sends alerts to the monitoring program when certain blades fail.
CompactPCI: The Standard for Blade Servers
The CompactPCI open standards architecture balances industry standards for hardware, operating systems, and application development tools to enable the rapid and effective development of high-margin value-added telecom services, while enabling the transformation of telecom construction traditionally based on proprietary hardware and software architectures, and the enjoyment of an open system. At the same time, it enables the transformation of telecommunications construction, which has traditionally been based on proprietary hardware and software architectures, to enjoy the benefits of an open system that brings significant cost reductions and a mass market industry standard operating system. This shift has enabled equipment and service providers to find millions of developers and begin adopting the highly reliable, scalable, and high-performance CompactPCI platform for broadband communications.
The CompactPCI bus standard is the foundation upon which blade servers are built. It is the only standard, and it is also the origin of the standard dispute. There are currently two major versions of CompactPCI, version 1.0 and version 2.0, which differ in the degree of refinement of the interface definition. Early blade servers all use CompactPCI 1.0 standard, the backplane bandwidth is also limited to 32-bit PCI, these products belong to the first generation of blade servers. 2002, the latest launch of blade servers part of the use of CompactPCI 2.0 standard, the backplane to support the 64-bit PCI communication, called the second generation of blade servers. Due to the different versions of the standard, the two generations of blade servers are not fully compatible with each other.
So far, only HP has claimed to design blade servers in full compliance with the CompactPCI standard, while other server vendors have only followed CompactPCI in terms of bus and interface standards, and have not fully followed the standard in terms of blade size.
Application Model Guide
Blade servers are used in a wide range of applications, especially for compute-intensive applications such as weather forecast modeling, data acquisition, data simulation, digital imaging design, aerodynamic modeling, and so on. And for industry applications, such as telecommunications, finance, IDC/ASP/ISP applications, cell phone base stations, video-on-demand, Web hosting operations and laboratory systems, blade servers can still make a big difference. The emergence of blade servers in the server market at the end of 2001 to occupy a piece of the market share is not small compared to rack-mounted servers. With the development of technology in 2002, especially InfiniBand technology began to play an important role, blade servers will gradually become mainstream servers and occupy a large market share.
Blade servers are used in a wide range of applications. Here we list two typical application models for a brief introduction.
Application Mode 1: Web Server for Websites
This approach takes full advantage of the high density, clustering, and remote management of blade servers. Web sites can use blade servers to form high-density clusters, used to achieve high access to the Web server, and then connected to the back end of the high-end servers or clustered system as a database server. Storage service providers can use the same front-end solution, and the back-end with NAS devices to provide storage services. The advantage of blade servers in this type of application is that they take up less rack space than regular rack servers, which can effectively save hosting costs.
Application Model 2: SMB Web Server
Today's enterprise networks are multifaceted and require a wide range of services, some of which can be installed on a single machine, while others require the use of at least one backup machine or cluster. In contrast, any one blade system can either run standalone or be clustered or backed up with other servers. Pairing is based on the actual needs of the organization. This approach can fully utilize the advantages of blade servers in terms of ease of management, flexible configuration and good scalability. The use of blade servers for clustering and combined with the storage area network, which can be capable of large data volume throughput of database parallel processing. For organizations, this high density not only saves valuable cabinet space, but also saves cabling costs and saves power, which reduces UPS requirements.