Virtualization Technology News and Information
Mainframe: Should It Stay or Should It Go?

Quoting ComputerWorld

An IT manager in a telecommunications company recently told me that he is waging a major battle over the mainframe. It seems that the high-level executives in IT (along with the company’s CIO) have been singing a familiar song: “The Mainframe Has Got to Go.” The execs are principally concerned with three things:

  • The costs associated with maintaining mainframe hardware and software. 
  • The long-term availability of packaged software currently running on the mainframe. 
  • The long-term availability of labor for mainframe applications.

This battle — or heated discussion — is taking place at many large companies today. Organizations that are running core business systems on mainframes are weighing when and if to migrate off of them.

If they do decide to migrate, these projects are guaranteed to be large, expensive and overwhelming from a business change perspective. Midrange systems vendors and software companies are stirring the pot by talking up the benefits of things like server virtualization and service-oriented architecture (SOA).

The telecom company IT manager is not in denial. He recognizes that the mainframe — eventually — will have to go. He just questions whether now is the right time. Here’s his response to the executives’ concerns. 

Certainly, mainframes are expensive when compared with midrange systems. However, the mainframe is all about economies of scale.

So once you’re using mainframes, adding more MIPS is actually fairly attractive from a cost point of view. That’s the case in this company. It owns one of the machines outright and has so many mainframe MIPS that the cost of adding more is quite competitive.

Many of the applications the company is running on the mainframe are Cobol-based; they’re used to do things such as customer billing. The money this company has already sunk into buying, tuning, tweaking and maintaining these applications runs well into the billions of dollars. Though these applications might not have been elegant initially, at this point and in this company, they are actually quite functional and stable.

To migrate these applications to a new platform would be a huge expense. It would also distract the business managers and users substantially. At the end of the migration, the application would look a lot like what they currently have.

Labor is probably the issue where there is the most disagreement. The executives are rightly concerned that the largely baby-boomer-based workforce that is expert on these platforms is retiring, and no viable replacement is on deck. My discussions with many large organizations suggest that this is true — this workforce is drying up, and companies are struggling to replace these staffers. The telecom IT manager argues well on this point, though. He suggests that the whole mainframe/Cobol skills crunch is more hype than reality. He says that if Cobol programmers were really in such great demand, we’d see their salaries going up.

By all estimations, such salaries are not rising dramatically. So salaries (the equivalent of price in an economic model) are static, demand is supposedly up, and supply is down? That scenario doesn’t make sense from an economic standpoint. One possible explanation: At this juncture, companies don’t perceive these skills as worthy of hot-skills pay. So Cobol specialists may be out there but unwilling to work for the salaries that are being offered.

On top of these arguments, my contact suggests that it’s never a great idea to be an early adopter of a new application architecture such as SOA. Best to let the market work out who the dominant players will be.

One could reasonably argue both sides of this dilemma. If your company is battling over the mainframe, I recommend you do the following:

Take a close look at the cost structure of the mainframe environment. Are you achieving economies of scale? Consider the costs of all elements, including acquiring new hardware, migrating software and consulting.

If your applications are custom-developed, your organization will be reliant on a skill base that is shrinking. Discuss with your human resources group the viability of hiring and retaining people with the necessary skills in the next five years.

If your applications are packages, find out from your vendors whether they plan to continue supporting the mainframe.

There’s no right or wrong answer here. Whether it’s time for the mainframe to go depends on many factors. Just be sure you understand how they play out in your company.

Read the original, here.

Published Saturday, October 07, 2006 11:22 AM by David Marshall
Filed under:
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
<October 2006>