Previous Articles >> EMC to bridge industry academia gap Build your own IP SAN

Related Articles >> HP intros new Integrity servers Manage and monitor your UNIX servers AMD selects HP blade servers

hey handle computer-intensive jobs, process data and forms and are the backbone of your 24/7 IT infrastructure. We are talking about servers, of course. In the computing ecosystem, servers are indeed the vital link. As a result, CIOs and IT managers today are looking at server buying decisions a lot more closely these days. They are looking at a lot of things such as cooling capacity, performance per watt, and interoperability. They also consider the RoI and technology. Today, servers are not seen as just boxes. They are now looked at as a vital resource that enables organization wide productivity.

A reliable IT architecture is only possible by deploying a mix-and-match of server technologies that can run a range of software applications. Any investment on servers is based on ramping up the backend IT infrastructure that can scale up according to the enterprise demands. Over the years, CIOs have come to focus on peak performance at affordable prices. Industry experts believe that the key to success in the server market lies in understanding the user demands and providing technologies that are affordable.

For instance if we look at the demands from the telecom vertical, which is a major consumer of all configuration of servers, the companies have looked at servers which are scalable and able to match the demands that increase by the day. Every cellular operator provides a range of value-added services and has a consumer portal through which the subscribers can do a range of tasks like checking the usage patterns to printing a copy of the latest bill. At the back end, the operator runs a plethora of applications and all these call for high-end servers. The vendors in turn have responded to these demands by way of carrier grade servers. Using these servers, telecom providers can deploy products and accrue greater benefits like lower TCO and greater RoI.

Yet another key trend over the year that saw escalation is that of running multiple OSs on a single server. This is mainly due to the multi-platform strategy enterprises adopted over the year, and the equal importance they gave to open source in addition to running proprietary software.

The x86 servers
Once an exclusive terrain of Intel, the x86 market dynamics is changing everyday. This is mainly due to the escalation of AMD's stature in the x86 space. The x86 architecture was first developed by Intel, and is one of the biggest server markets right now. The market acceptance for x86 is driven by two factors-reliability and affordability. This makes it a volumes segment in the server space as organizations have accrued distinct value by adopting x86 technology. On the affordability side, x86 server prices have been going southward and the distinction between a professional PC and a server is blurring.

One of the key trends seen over the year in the x86 server space is the rapid adoption of the 64-bit processors. The move from 32-bit to 64-bit computing offers a huge performance improvement, a dramatic increase in reliability, thus enabling enterprises to use computing resources in new ways.

Also, better hardware features that came with the new systems designed for these new processors resulted in much better performance over the 32-bit processors based systems. Even for the same set of 32-bit applications and 32-bit operating system, the performance on new systems was at least 20% better. Over the past one year, the adoption has reached high levels for the two-way and above x86 servers.

Unix: Business as Usual
The Indian server market's traction for Unix continues as RISC-based deployments from verticals such as telecom, BFSI and FMCG keep up the demand. Enterprises which looked at better scalability went in for Unix as they were able to scale their systems simply by adding more processors. This is a key consideration for IT departments running a data warehouse or decision support systems that quickly grow to multi-terabyte sizes. RISC servers also provide superior throughput due to their basic architecture, thus giving customers high operational efficiencies and stable performance. Reliability is the hallmark of Unix and enterprises whose emphasis is more on uptime and scalability, looked only at Unix. The excellent performance, scalability and ease of integration with Linux makes Unix a time-tested option in the server space.

The Unix business has become extremely competitive with the escalation of IBM in this space. The Unix market is a three horse race with Sun, HP and IBM battling it out.

Unix's reliability received a fillip with the recent Yankee Group/Sunbelt 2006 Server Reliability Survey. As per the survey, all server operating platforms ramped up in reliability, but Unix servers recorded the least amount of downtime. For instance, the leading Unix distributions-Solaris, AIX and HP UX-topped the server reliability charts as corporate Unix users reported experiencing only about 600 minutes of outage per server, per year. So clearly, Unix is the preferred choice for 24/7 computing facilitators like data centers.

Blades: Blazing Forward
The year gone by was a significant one for blade servers as they successfully crossed the cusp and became mainstream. As computing options continue to grow, the escalation of blades is commented on by industry experts based on the scale up vs scale out phenomena. For instance in scale up operations, one adds processors, memory and I/O to a system thereby increasing its computing power. In the event of scale out, the emphasis is on addition of multiple low-end systems in a clustered configuration for greater performance, availability and reliability of the final solution. Here is where blade servers come into play. Blade servers are an ideal way to physical consolidation of multiple heterogeneous servers. The blade servers category is the fastest growing segment in the server space.

One of the biggest advantages of blade servers lies in their power-saving feature. If we take a large data center environment, blades bring in definite cost advantages. For instance in a 1U rack mounted server, close to 30% of the power is consumed by the CPU, followed by memory at 11%, PCI buses at 3%, backplane at 4% and disk drives at 6%. The remaining 46% is consumed by the cooling components like fan and power supplies. In a blade environment, industry experts aver that the 46% power consumed for cooling in 1U servers comes down to just 10% as they use common cooling and power on a shared basis. This brings in twin benefits-less power consumption and better heat distribution. On the TCO front, a single blade costs 25% to 30% less than a 1U server and is 33% more effective in terms of power consumption and takes half the floor space.

A key driver of growth is cost performance. Today, justifying the cost of blade servers is less of a challenge for IT managers compared to the early days when blade servers were first introduced and were much more expensive. Another driver of growth is server consolidation. Blade servers are scalable and can help data centers save on real estate cost. The ease of installation and maintenance makes administration and support much easier.