IT Management

Every CIO has contemplated at some point replacing the mainframe with a completely server-based data center. It’s a no-brainer, right? The servers are so easy to use; they’re flexible, elegant, and virtualized. In addition, you no longer have to worry about where to find mainframe-experienced personnel. Ah, if only it were that simple …

There are many things a CIO must consider when mainframe and distributed server systems are present in an organization, including:

  • Mainframe vs. distributed server farm electricity and space costs
  • Quantity of mainframe personnel needed vs. distributed server farm personnel needed
  • Mainframe up-time and security vs. distributed server farm up-time and security
  • Mainframe TCO vs. distributed server farm TCO over a five-year period
  • The effectiveness of mainframe virtualization vs. distributed server farm virtualization
  • Oracle and other license fee savings of consolidating Linux onto the mainframe vs. virtualized distributed server consolidation
  • More advanced IT technologies also exist on today’s mainframes, so new workloads can be created just as fast as on distributed server farms.
  • Finding mainframe-skilled personnel vs. finding distributed server-skilled personnel.

While there are many considerations, all CIOs would be well-advised to quickly banish the thought of replacing the mainframe with all distributed servers because they would face a very unpleasant surprise in the future. Here’s why.  

The Total Cost of Operations (TCO) for a mainframe is 10 times less than most distributed server farms and requires far fewer people to efficiently maintain and operate. Today, there are many CIOs with one or more mainframe systems and a vast number of distributed systems in place. Mainframes are less expensive than their distributed counterparts in several ways, including electricity, floor space, and manpower. So, when you choose the distributed servers over a mainframe, you lose tremendous competitive advantage. Worse yet, such a decision isn’t easy to reverse.

Of course, both mainframe and distributed server groups will argue their system is the best solution. CIOs who possess a significant amount of IT experience and have been meeting business goals on a reliable and continuous basis for years have little difficulty determining the best IT solution for their organizations. These experienced CIOs understand the technical benefits of the mainframe as well as the best way to leverage this asset for maximum competitive advantage.

However, there also are CIOs whose experience is mainly on the business side of their organization. This group can face some yet unrealized, significant dangers in their decision-making when it comes to choosing the best IT solution for their organization. It’s easy for CIOs whose experience largely comes from the business side of an organization to make a computing platform decision that looks and feels correct only to discover five to 10 years down the road that their TCO has far exceeded their most liberal forecasts. Once there, they’re trapped in a sea of costs with no easy or inexpensive escape route.

But What About Being Able to Develop Applications Faster?

CIOs who focus on distributed platforms are able to quickly implement applications using Microsoft .NET, which is a developer favorite. These applications can quickly be developed, eliminating the symptoms of slow, methodical, development times. However, faster application development doesn’t yield the stable, methodical, wellgoverned IT department of mainframe computing environments that’s legendary for maximum security, availability, and dependability.

Figure 1 shows a set of calculations based on the WinterGreen Research ROI calculator, which contains a detailed analysis of the different factors impacting IT and data center costs. Different factors impact costs, including the number of distributed servers and MIPS needed to do the job, the N-way effect, workload variations, the level of processor-intensive computing to be done, and the variety and size of the workload to be processed. System calculations also depend on how well the software is optimized relative to the hardware platform on which it’s running.

2 Pages