Print

Comment

Email

Digg

Del.icio.us

Reddit

Twitter

In the early days of computing, mainframes ruled enterprises. Computing was still in its formative years and Moore's Law in its infancy. In those days, mainframe computers, also called the big iron, were on a pedestal so high that they were a symbol of power. More prestigious was the job of the mainframe programmer who commanded great salaries. But come 1990s computing models rapidly changed with the advent of servers that set the stage for distributed computing. During this time mainframes and their applications started becoming legacy as enterprises started migrating to new hardware configurations. Moreover, the operating systems landscape also changed, as a result enterprises acquired different operating systems and mainframes which just simply could not run multiple operating systems days.

A Peek into the Past
In the formative years, like in the 1960s, mainframes had one major drawback-there were no user interfaces. But in the 1970s, this flaw was rectified and mainframes had user interfaces. In the 1980s they also acquired graphical capabilities. The launch of the z series mainframes by IBM created a new premise. No longer were mainframes considered legacy, rather an integral part of the IT infrastructure of large enterprises. Today a mainframe can run diverse operating systems.

So, from the mid 1990s computing moved away from centralized to the distributed model and this is precisely the reason for the complexities many of the enterprises are facing today. For instance, many enterprises have acquired a good deal of servers distributed across the enterprise running multiple applications and operating systems. Due to this excessive distribution, IT managers are forced to consolidate their hardware in order to create a single view of their IT infrastructure. This move to centralization again is infusing fresh blood into the mainframe market. With consolidation and virtualization becoming a major thought augurs well for the mainframes as it is seen as the reliable hardware backbone that can offer a panacea to those companies struggling to manage compute intensive scenarios and have plethora of distributed servers.

Beyond Legacy
Mainframes are often associated with legacy apps. With the emergence of modern business apps like ERP, what are the issues? Moreover, how OS neutral and open they have become-like running UNIX, Linux, etc. Mainframes have traditionally supplied computing needs for CICS/IMS based applications, which are now considered legacy. Sreenath Chary, business unit executive, System z, IBM India, says, "The mainframe today runs Websphere the most used open platform. This makes the mainframe run Java natively under Websphere. In addition, mainframes also runs the latest Linux distributions, which makes it the most open platform that supports multiple operating systems as standard." Moreover, ERP applications from various vendors like SAP, Peoplesoft and others are all supported on the mainframe as well. These support the mainframe natively or support running on the mainframe via Linux or Websphere as multiple choices are available.

'We are aggressively pushing for the usage of mainframe into accounts in India with proof points of how mainframes could reduce the TCO in most instances' -Sreenath Chary, business unit executive, System z, IBM India

Need for Mainframes
Chary says, "Mainframes perform extremely well when there are contentions for data amongst a host of applications that need to run simultaneously. Its capability to provide consistent levels of services under such conditions is unparallel". Vendors like IBM, who dominate the mainframe space are bullish about their resurgence and escalation. This makes mainframes the perfect environment where there is increasing need for real-time database access and update, exploding transaction volumes and reliability to serve consistently.

The defined value of mainframes can be evaluated by their ability to churn out computing tasks in the most efficient manner. Most distributed servers are utilized to a lowly 15%-20% for Intel boxes and to about 30% for UNIX boxes. This is a huge wastage. Moreover, the cost of managing the complexity with hundreds of distributed servers is being felt more and more in the areas of cost of maintenance, cost of hardware failure, cost of upgrades, real estate space occupied, and lastly power consumption. Hence, to manage these complexities the panacea lies in virtualization through mainframes.

Page(s)   1  2