Print this article

Comment This

Email this article

Servers are the backbone of all the enterprise IT networks. Corporates run all of their critical applications on them.

Today servers run applications ranging from specialized tasks such as primary storage, Web hosting, telecommunications, file and print services to application, and database hosting etc. There has been a steady rise in the  demand for servers as more businesses got e-enabled and began transacting over the Internet. This has resulted in the evolution of Internet data centres, which are also called server farms and the expansion of enterprise networks. For buying a server, important aspects such as processors, brands and other technical specifications need to be considered.

 Today's technology behind servers has made them more reliable, faster and more efficient. They can  support data backup and security, reduce data bottlenecks for easy information flow and are scalable. Today, servers have become an indispensable part of an enterprise network

ENTRY-LEVEL SERVERS
Entry-level servers are Based on the x86/x64/UNIX architecture, ranging from single processor systems to up to 4 processors. For a benchmark, IDC defines entry-level servers as those that sell for less than $100,000 -a broad group that includes RISC (reduced instruction set computer), 32-bit Intel and PowerPC-based servers; blade servers; and 64-bit Intel Itaniums. These servers are able to support significant internal expansion of hardware and enterprise can deploy most of these server types to provide their computing infrastructure.  

Entry-level servers are evolving from an unmanaged, single-CPU server to a more powerful, multiple-processing, highly available, and fault-tolerant machine. Latest development in server technology for these servers have made them more compatible  to enterprise needs.

Powerful features, such as memory mirroring formerly used in midrange and high-end Unix servers, is migrating down as enterprise demand more value for money. The compute power of an entry-level server matches what a midrange would have offered four or five years ago. These servers, today, are capable of providing outstanding performance in Web surfing and database transaction processing and simultaneously cut costs by low power consumption. These technologies include first failure data capture for, disaster recovery and other relating contingencies along with operating system flexibility. They offer the scalability, performance, reliability, and security desired to meet the demands of e-business workloads. Ample storage capacity enables enterprises to run powerful applications without having to buy additional attached storage devices. Wireless administration is another feature of these servers, which allows system administrators to manage it remotely using a handheld wireless PDA.

	EXPERTS PANEL
	Amod Phadke, GM (enterprise business), Gigabyte Technology India Avijit Basu, country manager (enterprise servers and storage), HP India Unnikrishnan KP, marketing director, Sun Micro Systems India

High-end memory features such as partitioning capability, more powerful and faster processors, hot-swappable memory, disk drives, and PCI (peripheral component interconnect) adapters are becoming standard features on servers as users require more capability for less money. Features such as hot-pluggable power supplies and dynamic de-allocation of processors, chipkill memory, and bit-steering that help lower the number of memory failures are becoming standard options for an entry level server.

Entry-level datacenter class servers are economical and provide a secure and flexible platform for expansion along with excellent serviceability and reliability. These servers offer outstanding price/performance and are designed to deliver horizontally scalable services and solutions. The application areas of these servers have also evolved. Today they can be used for running server applications ranging from file and print sharing, application development, network security and management, customer relationship management, enterprise resource planning to
e-commerce etc. Applications such as e-mail or databases, Web services, and security applications are being run on them because these servers are reliable and durable.

MID-RANGED SERVERS
Mid-ranged servers are based on x86/x64/UNIX - Designed for compute density backed by enterprise-class features; 4–24 processors, 8–48 threads. These servers offer full-speed connectivity with multiple GB Ethernet LANs, and provide redundancy in major components for high availability and inclusion in clusters. Small to midsize enterprise can leverage excellent performance from powerful midrange server products. Midrange servers are priced at $25,000 to $1 mn.

Mid range servers offer a lights-out service processor and software for provisioning and managing common applications. They also have partitioning capabilities, which provides them with built-in flexibility to perform a range of functions.

Midrange servers are ideal for those enterprises, which are looking at consolidation of their application and computing resources. If the focus is on cutting costs and buying servers just to run current applications, enterprises should invest strategically in these servers to handle future workloads effectively. Features such as virtual partitioning and self- repair capabilities make the mid range servers reliable and highly secure systems that simplify IT, while reducing costs.

Introduction of multicore technology and processors that extend the functionality of 32-bit operating environments and applications has revolutionized the midrange server market.

HIGH END SERVERS

High-end servers are Massively scalable, highly available servers designed for mission-critical network computing; 36–72 processors, 72–144 threads. With the advent of decent SAN solutions high end servers are gaining popularity steadily They are now able to offload a lot of the storage overheads and can get down to their primary function of high speed data processing. High end servers or Mainframes are large and 'expensive' computers used mainly by government institutions and large companies for mission critical applications. They have hot-swapable power supplies, and uninterrupted power supplies all built in to the machine. Most of these machines run operating system such as UNICOS, OS/390, z/OS, or AIX UNIX. All mainframe operating systems are some derivative of UNIX compiled to support the extremely powerful architecture of these machines. 

CLUSTER SERVERS
CIOs today have one major objective and that is to convert the IT department from being a cost centre to a valuable business component of an enterprise.

Today's e-business infrastructure requires information technology solutions that can meet ever-increasing demands with high reliability and ease of management. Enterprises want solutions that offer the scalability to grow with the business, the flexibility to rapidly develop and deploy new services, and are economic also. Clustering is the answer to such needs of the enterprises.

Clustering is now being used increasingly as a cost-effective way to provide high-performance, high availability computing for a wide variety of commercial workloads. Applications such as business intelligence, engineering design, financial analysis, digital media and petroleum exploration are using clustered servers to achieve desired level of efficiency and performance.  

Clustering is the practice of connecting multiple processors or servers to cooperate on complex workloads as a single, unified computing resource. Because it behaves like a single large resource, a clustered system offers many valuable benefits to a modern e-business environment. Clustered systems combine the power of multiple servers to provide high processing capacity and can tackle large and complex workloads. Since the processing is spread across multiple machines, clustered systems are highly fault-tolerant: if one system fails, the others keep working.They can also span multiple geographic sites so even if an entire site falls victim to any disaster such as power failure the remote machines keep working. Clustering enables horizontal and vertical scalability without downtime. This allows incorporating additional processing power without interrupting business operations.

Companies are looking at ways to cluster smaller, low-end servers to achieve performance and reliability that is equal to or better than expensive, high-end boxes. Improved performance and reliability of low-end servers has given IT managers an absolutely new perspective of looking at clustering. IT managers are forced to do more with less in recent years and they have begun buying more low-end, but more powerful servers. Enterprises can expand horizontally and vertically adding more horsepower to the current servers and more servers to the configuration.   

Recently, cluster technology has grown to take a significant share of the technical computing market. Enterprise user prefers to exploit the opportunity of getting higher performance by combining multiple servers rather than build or buy powerful special-purpose servers. Clusters are expected to currently account for more than half of all technical computer revenue.  The rapid adoption of cluster technology is beneficiary for an enterprise because clusters leverage improvements in industry-standard technologies thus, resulting in exponential increases in component capability without any cost. They are highly cost-effective in capacity computing environments.  Clusters are one of the most cost-effective solutions to problems with highly parallel applications. The ability to quickly and easily add nodes to clusters provides a  simple  method  to  increase  capacity  in  small increments. Thus, clusters can help enterprise to manage the IT infrastructure efficiently and effectively.

BLADE SERVERS
Blade servers can be defined as servers on a card. The strength of a blade server is in the elements they share, which include the shared data pathway. These servers enchance an enterprise's cost effectiveness as they save valuable floor space, reduce the amount of cabling, and provide great management flexibility. It also enables great scaling: Blades can be scaled simply by adding cards to the existing chassis. The whole benefit of blades is that the network elements can put side by side like books in a bookshelf.

Blade servers are gaining popularity in India and around the world. The key driver of blade has been the driving down of total cost of ownership (TCO), which is a major issue to the CIO and CTO. The reduction in cabling combined with the simplicity of adding servers has been one of the main drivers for companies with more than 25 servers to deploy blade servers.  

Blades offer an economic and more reliable way to handle massive computations based on parallel processing. It's an economic solution with respect to space and energy consumption.

Blades are proving to be a boon for the industry, a technology that saves space, boosts reliability, streamlines management, and ultimately, cuts costs.

Single servers can't match the degree of efficiency that blades can achieve by sharing a common high- speed bus and U-mounted support component. Along with letting enterprises take advantage of the efficiency and reliability of clustered technology, shrinking a server to the size of an add-in card permits impressive densities. Blade servers offer between 3–10 times the density of conventional servers, along with substantial improvements in the management and systems integration cost. They are suitable replacements for rack-mount and standalone servers that are used in several tasks-from the network edge to application servers. They can also be configured to include load-balancing and failover capabilities. Individual blades are easy to swap out a board with a new one in the event of system failure as they are hot-pluggable. Moreover, placing servers near each other and managing them under a single application can streamline administration.

New enhancements have been introduced for blade servers in 2006 for boosting its networking features and flexibility. They include forward and backward compatible chassis, expanded chasis, power PC chip, Intel's Itanium processor etc. These will enable some blade servers transfer data at a much higher speed. The speed boost comes in time to meet customers' demands for consolidating the workloads from several machines into one to reduce data-center clutter and bring down infrastructure and power-consumption costs.

Blade servers are a fast-growing part of the overall server market. IBM's BladeCenter leads the market with 42% of the $419 mn spent on blade servers in the second quarter, according to Gartner. The significance of blade servers should increase, though, with sales of $9 bn in 2008–29% of the entire server market, IDC forecasts

The chassis into which the blade server slides, simplifies the snarl of power, management and network cables that protrude from standalone servers, and it supplies shared infrastructure such as network switches and power supplies.

Blades are well suited to run business-critical applications and have the potential of becoming an important building block for IT infrastructure in the future.

Blade servers will benefit Service providers such as Internet Data Centres (IDCs), telcos and large enterprise data centres tremendously. The use of blade servers will provide modular and flexible capabilities and help organisations respond quickly to changing business conditions in a cost-efficient manner.

Blades have been adopted by Indian enterprises and still others are in the process of evaluating this option. With the increase in complexity of operations and resource necessities of new enterprise applications, blades will look increasingly attractive to CIOs who want to upgrade and consolidate their server architecture, particularly in the Web/edge tier.  

SERVER VIRTUALIZATION
The latest buzz in the server technology is server virtualization. It is the masking of server resources from server users. The rationale behind the technology is to spare the user from having to understand and manage complicated details of server resources during the up scaling and maintainance. Server virtualization allows an enterprise to 'scale-out' with a number of smaller computers to address its growing processing needs rather than having to buy a more expensive multi-processor computer. Server virtualization enables all developers to share the same machine during developing or testing new programs; provides the ability to support already existing legacy applications with the new ones and the ability to simulate distributed network applications on the same physical server.

Server virtualisation can be  utilized to solve traditionally difficult problems such as security because segmentation  of resources can help in boosting serurity. With server virtualization, segmentation can be accomplished by executing software processes on logically separate virtual servers instead of the traditional way of doing it at network level. Thus, a single server compromise may not impact the system as a whole. erver virtualization also provides a cost-effective means validation of a system patch or upgrade within a test environment prior to production deployment of duplicating or imitating a large production environment. The virtual environment can then be used to perform validation activities. As virtual server technologies unite with storage technologies (such as cloning, snapshots and shadow copies), traditional difficulties with backup and recovery scenarios may be eliminated. It also makes the process of packaging remote test environments economically feasible, thus the quality of software should naturally rise and more tasks may be performed offshore.

MARKET INFORMATION
For the first time, the server market in India is expected to cross the 100,000-unit mark in 2006, according to IDC. The milestone will enable India to supersede Australia as the second-largest server market in the Asia Pacific (excluding Japan) region in terms of unit shipments. The big blue retained the number one spot in the worldwide server systems market with 32.3% market share in factory revenue while HP stood at the number two spot in terms of factory revenue with 27.8% share. However in terms of unit shipments, HP came in at number one position worldwide with 28.8% server shipment share.

Sonia Sharma 
sonias@cybermedia.co.in