Blade Server

1. Introduction
Server is an adjective in the term server operating system. A server operating system is intended, enabled, or better able to run server applications. A server computer (often called server for short) is a computer system that has been designated for running a specific server application or applications. A computer that is designated for only one server application is often named for that application. For example, when Apache HTTP Server (software) is a company's web server, the computer running it is also called the web server. Server applications can be divided among server computers over an extreme range, depending upon the workload. Under light loading, every server application can run concurrently on a single computer.
 Server hardware
 Server operating systems
 Servers on the Internet
1.1 Server hardware
CPU speeds are far less critical for many servers than they are for many desktops. Not only are typical server tasks likely to be delayed more by I/O requests than processor requirements, but the lack of any graphical user interface (GUI) in many servers frees up very large amounts of processing power for other tasks, making the overall processor power requirement lower. If a great deal of processing power is required in a server, there is a tendency to add more CPUs rather than increase the speed of a single CPU, again for reasons of reliability and redundancy. The major difference between servers and desktop computers is not in the hardware but in the software. Servers often run operating systems that are designed specifically for use in servers
1.2 Server operating systems
The Microsoft Windows operating system is predominant among desktop computers, but in the world of servers, the most popular operating systems—such as FreeBSD, Solaris, and GNU/Linux—are derived from or similar to the UNIX operating system. UNIX was originally a minicomputer operating system, and as servers gradually replaced traditional minicomputers, UNIX was a logical and efficient choice of operating system for the servers
1.3 Servers on the Internet
Almost the entire structure of the Internet is based upon a client-server model. Many millions of servers are connected to the Internet and run continuously throughout the world.Among the many services provided by Internet servers are: the Web; the Domain Name System; electronic mail; file transfer; instant messaging; streaming audio and video, online gaming, and countless others. Virtually every action taken by an ordinary Internet user requires one or more interactions with one or more servers.
Connects to:
• Internet via one of
 Ethernet
 Modem
Common Manufacturers:
 International Business Machines
 Sun Microsystems
 HP
 Apple Computer
 Motorola
 Dell
 Supermicro


2. History
Complete microcomputers were placed on cards and packaged in standard 19-inch racks in the 1970s soon after the introduction of 8-bit microprocessors. This architecture was used in the industrial process control industry as an alternative to minicomputer control systems. Programs were stored in EPROM on early models and were limited to a single function with a small realtime executive.
The VMEBus architecture (ca. 1981) defined a computer interface which included implementation of a board-level computer that was installed in a chassis backplane with multiple slots for pluggable boards that provide I/O, memory, or additional computing. The PCI Industrial Computer Manufacturers Group PICMG developed a chassis/blade structure for the then emerging Peripheral Component Interconnect bus PCI which is called CompactPCI. Common among these chassis based computers was the fact that the entire chassis was a single system. While a chassis might include multiple computing elements to provide the desired level of performance and redundancy, there was always one board in charge, one master board coordinating the operation of the entire system. PICMG expanded the CompactPCI specification with the use of standard Ethernet connectivity between boards across the backplane. The PICMG 2.16 CompactPCI Packet Switching Backplane specification was adopted in Sept 2001 (PICMG specifications). This provided the first open architecture for a multi-server chassis. PICMG followed with the larger and more feature rich AdvancedTCA specification targeting the telecom industry's need for a high availability and dense computing platform with extended product life (10+ years). While AdvancedTCA system and board pricing is typically higher than blade servers, AdvancedTCA suppliers claim that low operating expenses and total cost of ownership can make AdvancedTCA-based solutions a cost effective alternative for many building blocks of the next generation telecom network.
The name blade server appeared when a card included the processor, memory, I/O and non-volatile program storage (flash memory or small hard disk(s)). This allowed a complete server, with its operating system and applications, to be packaged on a single card / board / blade. These blades could then operate independently within a common chassis, doing the work of multiple separate server boxes more efficiently. Less space consumption is the most obvious benefit of this packaging, but additional efficiency benefits have become clear in power, cooling, management, and networking due to the pooling or sharing of common infrastructure to supports the entire chassis, rather than providing each of these on a per server box basis.
[One opinion is that…] The architecture of blade servers is expected to move closer to mainframe architectures. Although current systems act as a cluster of independent computers, future systems may add resource virtualization and higher levels of integration with the operating system to increase reliability.
[An alternate opinion is that…] The architecture of blade servers will remain a set of separate servers within a common chassis. But, the chassis will be offered in a wider variety of sizes, providing different levels of aggregation and different cost points. On top of this independent server blade structure, management capabilities will be enhanced that allow operation of a collection of servers as a single compute resource pool. The administrators and users will not need to care which server an application is on, since the management software will automatically take care of allocating server (and I/O) resources to the application as needed. Over time, the individual servers will not need to be uniquely managed, but the overall roubustness of the data center will still benefit from the separation of individual servers, whose interconnect is provided by standard networking technology. This provides isolation at a very low level so that a fault in one piece of hardware impacts only that work executing on that specific hardware at that time. The broader management software can then recover the impacted work by initiating it on another server blade elsewhere in the data center.
3. Overview
It is now estimated that after just three years on the market, blade servers account for seven percent of all server shipments sold; with estimates showing that by 2008 they will make up thirty percent of all servers sold. It is expected that they will be the fastest growing server form factor through 2009. There are many factors to consider when deciding if blade servers are right for your organization. These include but are not limited to costs, work loads and how predictable the work load is, performance and power consumption, uptime and repair, proprietary designs and disk storage considerations. With 3rd generation blade servers on the market the jury is still out on whether they will truly fulfill their promise of resiliency, repair efficiency, cost efficiency and dynamic load handling. Blades must continue to evolve toward this vision while still offering improvements in modularity, back-plane performance and the aggregation of blades into a single virtual-server image.
Currently five manufacturers make up more than 75% of all the blade servers sold in the market. These manufacturers are HP, IBM, Sun, Fujitsu/ Siemens and Dell. With the list of target applications growing for blade servers and the number of manufacturers increasing the future for this technology certainly looks bright.

4. Why Blade Servers?
The provisioning capability of our existing rack-mounted server environment is both lengthy and inflexible. Our search for a better approach led us to blade servers, an emerging technology that promised reduced costs in areas such as capital expenditures, operational expenses, and physical plant requirements while improving efficiency. Blade servers allow for up-front provisioning of the chassis and switch components when adding compute blades as needed. This provides a more dynamic provisioning model, which results in just-intime provisioning as well as the ability to work around data center freeze periods

5. What are Blade Servers?

Blade servers are a new trend in data center technology. Hardware manufacturers combine all server system hardware—one or more microprocessors, memory, disk drives, and network controllers—onto a single electronic circuit board, or blade. The blades are designed to plug into a system chassis and share common components such as power supplies, fans, CD-ROM and floppy drives, Ethernet switches, and system ports. Blade servers are sometimes referred to as headless devices because they have no monitor or keyboard of their own.
Individual blade servers come in various heights, including 5.25 inches (3U), 1.75 inches (1U), and smaller. (A U is a standard measure of vertical height in an equipment cabinet and is equal to1.75 inches.) The thin, modular design of these blades allows more computing power to exist in a smaller space than typical 1U rack servers.
Blades servers can help cut network administration costs and improve server manageability by consolidating many widely distributed servers onto one rack. This decreases the amount of space needed in a data center and eliminates the need to travel between server locations. The shared networking and power infrastructure reduces the number of cables coming off of a rack and lowers the chance of an administrator accidentally unplugging the wrong device. The reduced heat output and power consumption can also help trim energy costs. Another advantage of blade deployment is the ease and speed with which you can add more servers. Rather than spending hours installing a rack-mounting rail kit, administrators can plug in a new blade server in seconds. Replacing a damaged server is just as quick. Each blade is completely independent of the others; inserting or removing a blade has no effect on any other operational blade in the same chassis. Blade servers are ideal for organizations with large data centers and for those that use server farms to host Web sites. Blade servers are typically used in a server cluster that is dedicated to a single application (such as file sharing, Web page serving and caching, and streaming audio and video content). In a clustered environment, blade servers can be set up to perform load balancing and provide failover capabilities.



6. Architecture

A general blade server architecture is shown in Figure 1. The hardware components of a blade server are the switch blade, chassis (with fans, temperature sensors, etc), and multiple compute blades. The outside world connects through the rear of the chassis to a switch card in the blade server. The switch card is provisioned to distribute packets to blades within the blade server. All these components are wrapped together with network management system software provided by the blade server vendor. The specifics on the blade server architecture vary from vendor to vendor

7. Advantages of blade servers

Blade servers make it possible to accommodate a large number of servers and switches in one chassis at high density. The hardware unit containing a chassis is called a server blade, and the server blades are connected to switches called switch blades through a middle plane. Compared to traditional servers, blade servers have the following advantages
:
1. More servers can be placed in each rack.
2. They can accommodate computing devices and network devices in one chassis.
3. Switches and servers do not need to be connected by cables.
4. Server blades and switch blades are hot-pluggable.
5. They can accommodate equipment called management blades that allow administrators to collectively manage the entire hardware, for example, server blades, switch blades, power units, and fans, in a chassis.

7.1 Management Functions:

Management blades provide the following management functions:
• Configuration management
Manages the configuration and state of the blades in a chassis and detects changes that occur in the blades.
• Fault management
Detects faults occurring in server blades,switch blades, power units, and fans. This function also issues alarms to management tools using Simple Network Management Protocol (SNMP) traps.
• Power management
Performs on/off switching and rebooting of a chassis, server blades, and switch blades.
Figure 1 shows an example hardware configuration of a blade server

8. Issues
Business systems that use blade servers have a large number of server blades, and when a large number are in use, it is important to ensure they do not degrade system reliability. Although many factors can degrade a system’s reliability, the most important ones to consider are:

1) Hardware faults
2) Software faults
3) Insufficient resources to deal with unexpected load increases
4) Human error
Regarding factors 1) and 2), it is expensive to improve the quality of hardware and software. A blade server helps to reduce the service downtime, because new servers can be added and failed servers can be replaced without disconnecting and reconnecting the complicated network cables. Regarding factor 3), it is difficult to accurately estimate system size in an open system.
Moreover, in a Web system, it is difficult to accurately predict the amount of transactions. Therefore, it is necessary to build a system that includes an allowance for peak workloads. As a result, there are many resources that are usually not used and systems inevitably go down when the number of transactions exceeds the predicted level. On the other hand, the number of active servers in a blade computing system can easily be increased, so only the minimum number of servers needs to be used in the initial installation and servers can easily be added when the system load increases. Regarding factor 4), it is impossible to eliminate human error completely, but the risk of it occurring can be reduced by automating system operation as much as possible. However, in a traditional system, servers and networks are separated, so the system configuration strongly depends on human operation. With a blade computing system, networks and servers are consolidated in a single chassis, so it is easier to automate system operation and therefore minimize human error than in a traditional system.





9. Systems using blade servers

A high-availability and cost-effective system using blade servers is constructed as follows:
1. Bare-metal server blades are prepared for a server pool that can be shared by several business systems.
2. If a fault occurs, the environment and settings are restored to another server blade in the server pool.
3. When a server resource shortage occurs in the load sharing server group because of a load increase, a server from the server pool is added to the load sharing server group. Then, the environment settings are restored to extend the load sharing server group.
4. When the fault is repaired or the system load returns to normal, the server that was replaced or added is returned to the server pool.


Figure 2 shows an image of a high-availability system using blade servers. Provisioning provides a quick and flexible service by preparing IT resources in advance.
It brings the following benefits to customers:
 It enables customers to visualize the impact of physical faults on services and thereby reduce the time needed to recover from faults.
 It quickly reconstructs a system when a fault occurs and therefore reduces system down-time.
 It quickly adds servers when the system load increases and therefore reduces the amount of time the service level is degraded and makes it easier to use server resources more effectively.
 By automating intervention tasks such as installation and maintenance, it reduces the risk of human error and makes operation more cost-effective.

Management components
The management components are implemented by the blade server hardware and the existing management software. Table 1 shows the management functions required for provisioning on blade servers.


10. Comparison
10.1 Blade Servers

 Power consumption reduced from 260 amps to 60
 Heat generation reduced by 15%
 Total environment cost down 10%
 Proprietary
 Cheaper than rack after purchasing 8 or 10 blades
10.2 Standard Rack Servers

 Power consumption reduced from 260 amps to 220
 Same heat generation
 Total environment cost remains the same



11. Major components of the xSeries 336 Type 8837 server




 Light Path LEDs and buttons:
The Light Path LEDs and buttons are on top of the operator information panel. The following illustration shows the LEDs on the Light Path Diagnostics panel, followed by a description of the buttons and each LED.







12. Future Developments
At present, an administrator must analyze the root cause and make decisions to take action before provisioning. However, if human intervention
remains necessary, it is difficult to respond quickly to business demands, which change at dizzying speed. Moreover, it is difficult to take action in advance before a problem occurs.In the future, systems themselves will perform analyses,make decisions autonomously based on business requirements, and reconstruct systems organically. The key technologies for achieving these functions are considered to be system composition and autonomic control.
12.1 System composition
Designs logical system layouts and generates suitable physical arrangements using resource management. Its goal is to design a system based on business requirements (e.g., reduced system cost) and a service level agreement (SLA) to provide,for example, a guaranteed maximum response time.
12.2 Autonomic control
Automates provisioning tasks when a problem occurs. It predicts the future service level from the viewpoint of business requirements (e.g.,load increases in a specific period) and logs data of the past (e.g., the amount of resources used). If it predicts an inability to satisfy the SLA, it automatically performs dynamic system reconstruction. If the technologies and resource management described in this paper work together, autonomic provisioning can be realized. Figure 7 shows the relationships among these components.





Conclusion
Blade servers can greatly improve the reliability of business systems. Provisioning realizes swift action and efficient resource utilization when faults and load increases occur. In order to realize provisioning, resource management for configuration management, pool management, and automation is indispensable.In the future, provisioning solutions will have extended support ranges and these new solutions will form the basis of autonomic provisioning.





BIBLIOGRAPHY

 www.ibm.com/systems/bladecenter
 www.intel.com/products/server/blades/index.htm
 www.spectrum.ieee.org/apr06/1106/4
 www.sun.com/servers/blades/8000/
 www.hp.com/products1/servers/carrier_grade/products/atca_bladesystem/index.html