Articles
News
Jobs
Events
Store
Subscribe
- Newsletters
  
  The best articles, topics, & offers delivered straight to your inbox.
  
  Subscribe Now →
- Enterprise Executive
  
  The Magazine for IT Managers in the Worlds Largest Multi-Platform Enterprises.
  
  Get a FREE Subscription →
- Enterprise Tech Journal
  
  The Magazine for IT Staff in the Worlds Largest Hybrid Data Centers.
  
  Get a FREE Subscription →
- - RSS
  - LinkedIn
  - Twitter

Jul 10 ’12

It’s Time to Get Politics Out of the Data Center

by Bob Thomas

I realize that we’re getting into the heart of the 2012 Presidential campaign season and I enjoy a political fight as much as anybody. But I hate to see important IT decisions being made based on politics, religion or anything else that has nothing whatsoever to do with actual business-case issues.

Too many times I have seen and heard of decisions made to purchase hardware and/or software that were based not on what was actually best for the company making the acquisitions, but on the personal biases of the one making the recommendation or the final decision. It's time to get platform fiefdoms and politics out of the data center.

This issue came up again very recently while I was attending the IBM Edge2012 storage conference in Orlando. One of the attendees, an IT professional from a healthcare company with a multi-platform data center, stopped by our booth and in the course of our conversation I learned that for some reason this company had enough credits with IBM that they were given a zBX at no charge. Sounded like a pretty good deal to me.

So I asked the next logical question (always on the lookout for a potentially good case study to publish): How is the new zBX being utilized? The answer floored me. I was told that the zBX is not being used at all. Just sitting in a corner and not plugged in. But when I asked “Why?” the answer blew me away. It was because the folks who worked on the distributed systems side did not want any distributed systems to be hosted in that z/BX for fear of losing control to the mainframe side. Wow; makes you wonder who pays their salaries: their employer or the distributed systems vendors!

This little anecdote totally makes my point. Their decision was purely a political, personal decision; not one based on what was best for their company and the users. Now I fully understand that these types of decisions also frequently happen in the reverse where someone with a mainframe bias makes a similar irrational decision. Basically, for every workload there is an optimum place where it should be run based on the circumstances at the time, not personal prejudices.

In a lot of cases, the justification for a particular dogma is all about framing the issue, which the people doing it don't see as religion, but merely enlightened foresight. Suppose someone believes that mainframes will be an abandoned technology in xx years. I'm sure they can find some analysts to back this assertion up. Then the need to spend 1B$ or 10B$ can be framed as a matter of business survival and it's no longer a pure A versus B TCO debate. You can see this at work as Sun Solaris went from being "the new mainframe" to "abandon ship" in a very short period - what changed? Simply people's belief in the relative longevity of the systems.

At a reasonably large place I once knew, the effort to build a distributed system to handle a new financial system was determined by dogma to have to be "modern" and therefore a priori non-mainframe. A small vendor package was selected for >100x scale-up. When it had become apparent that the basic algorithms would not scale, and therefore the I/O was unsustainable, the entire team was sat down for a lecture directly telling us that making it work was "merely an effort of will". To illustrate the point the project leader (SVP) read to the assembled officers from "The Little Engine That Could" - and, no, I am not making this up! This is a belief system - aka religion/dogma - in its most obvious form, including parables and the idea that there are heretics to be shamed into recanting or cast out. I know of examples that are less starkly seen, but essentially similar logic. "COBOL is a dead language that no-one learns" -> "so we need to choose a system written in C or Java" -> "because the only package vendor (who uses C/Java rather than COBOL) doesn't support z, we will move this payments system to x86". Bang! - xxM$ gone.

The point is certainly not that it's "mainframe über alles". But how you look at and evaluate options is often gated more by how you see the future as where you are right now. It is quicker and it is simpler to apply this type of absolutist lens because it cuts off debate.

Posted by TheLittleEngineThatCouldnt on 2012-07-21 10:00:27

I appreciate the comments that have been written below. Fiefdoms in the data center based on personal prejudices are not conducive to efficiency, productivity, or a harmonious atmosphere and should be eliminated by courageous CIOs before they become too rigid.

I just heard of a very large company that is determined to get rid of their mainframes. At an estimated cost in excess of $1 Billion, the project is reportedly well behind schedule. My guess is that mainframes will continue to be in use there for years to come. I'd love to know how this decision was justified. Could it be that it was based on personal prejudices rather than on a sound business strategy?

Posted by Bob Thomas on 2012-07-19 11:49:53

Bob, as others said, no surprise. But people make decisions based on what they have been taught and what they feel comfortable with. Whole generations have not been educated on anything mainframe while others (or maybe even the same people) are slowly realising that a mainframe that is reliable, secure and efficient really DOES pose a threat to some parts of the distributed environment.

We should however remind ourselves of how we acted when people tried to move applications off the mainframe to distributed environments. I remember some colleagues who demonstrated the same behavior as we see now from our distributed colleagues.

We are all expected to do what's best for the company, but unfortunately, personal fears, ambitions and ego's often get in the way....

Posted by Marcel den Hartog on 2012-07-18 04:24:38

I had to laugh at this article based on a case of deja vu. Several years ago I was the data center manager for a small midwestern telecom company. We operated both mainframe (IBM MVS) and AS/400. Due to the bias of the IT chair, we were to not even consider any products from CA. At the time (the late 90's) it was almost impossible to get any practical use from a mainframe system without any CA software. The system sat, for all intense and purposes, cold in the corner never hitting a CPU percentate over 3% while the AS/400 systems (we had 4, and soon 6) were hammered mercilessly.

Posted by Paul King on 2012-07-16 07:26:36

Bob:

Several questions come to mind.

1.. Why did the company accept the box if it wasn't going to use it? Something that expensive is not something I would want in my company if it was unused and my company would be responsible if anything happened to it.

2. Since the z/BX is an extension to the Z196/Z114, was it for Intel/AMD or Power7?

3. Was the box usable to do something without additional hardware or software?

In short the mere acceptance of the box had political implications and whoever accepted it was playing poor politics. The design of this box as I understand it is to allow the running of Linux, Unix or Windows applications on native hardware (Intel/AMD or Power 7) with closer integration with System z. This would require at leas some effort on the part of some group even to test the theory for any given shop. This has a cost. Even if all of the hardware and supporting software either came with the box or were movable from existing racks, this is not something done over a lunch hour at a pizza party. All companies have politics if they are run by human beings. Since a z/BX must of necessity involve the distributed team, I have to ask why they weren't in on this from the time IBM made the offer. This sounds like something that someone just tried to dump on their laps. I almost have the feeling from the brief description, they were put in a bad position or at least made to feel that way. The z/BX is new, probably bleeding edge technology and I can see not wanting to add something like that to my plate without careful thought on the costs and benefits.

I know that when I was a systems programmer, I liked the idea of my installation being able to be a beta test site for software we wanted but I couldn't justify the risk and time commitment for the limited benefit it would have bought my company so I never brought up or pursued the idea. New things are nice but they have to have potential benefits that match the potential costs.

Posted by Clark F. Morris on 2012-07-14 19:04:49

Bob, I'm sure this doesn't come as a surprise to many of your readers. It's an all too common occurance and as you pointed out, one that is often supported by the management team for politcal reasons. They even do it with such impunity, that they don't mind openly discussing it with others. Distributed systems demand more people on staff to support them. This means a larger fiefdom and budget to manage. What's just as ironic, the vendor takes credit for another installed zBX.... even if it's turned off. A win-win for everyone but the true zEnterprize warriors; the Linux on z/VM types.

Posted by Len on 2012-07-13 13:55:39