Standardization Attention Deficit Disorder (ADD) or (SADD): (n) A condition in which one professes to support standardization, yet can’t help but be distracted by the newest, shiniest object – Opex costs be damned.

I’m a hypocrite. There. I said it. It’s almost therapeutic. Are you one too?

Here’s how I see it. We are all taking part in a great conspiracy. Many of us are both victors and victims in this circular history that we can’t help but repeat. End user organizations are spending way too much on IT services, and we are all at fault. Why? Let’s start with complexity. Every management vendor wants complexity. The more complex the environment, the more software and professional services they can sell. Hardware vendors? Ditto. Startups challenging incumbents? You got it. If you standardized you wouldn’t buy their products and they’d be out of business. Again – the newest, shiniest object is better than what you already have. And it’s cheaper! What’s not to love? IT pros generally like complexity at times because it lets us flex our intellectual muscle and show our value. Consultants and analysts? Check and check. Complexity equates to a greater need for advisory services.

Are we all a part of one of the greatest con jobs in history? Sometimes it feels that way. We can always find a “business reason” to make things harder than they need be. Or maybe we’re the victims? We’re being duped by a community that professes the values of standardization but on the other hand goes at length to justify anything but standardized approaches to IT challenges. Many of us suffer from SADD. So how on earth can a highly standardized approach to delivering IT services hold our attention? If a lack of standardization costs the business more money long term, then so be it.

We need to pit vendors against each other. Right? Group think often implies that it’s a better strategy. But who is it really better for? Vendors? Consultants? Analysts? The IT department? How often do we ever wonder if it’s best for the business? I’d argue not nearly enough. After all, in the quest to save 10-20% on one solution, what are you paying for new consulting, advisory services, and management products to deal with the added complexity?

A few years ago I blogged about emerging cloud technologies and what I called “the Wal-Martification of IT,” stating:

Think of public cloud providers as the neighborhood Wal-Mart. In many towns across the US, small businesses were swallowed by Wal-Mart. Many of these businesses were unwilling or unable to change their existing business processes or target markets in the wake of Wal-Mart’s entrance to their community. At the same time, Wal-Mart doesn’t exist in ghost towns. Look around most Wal-Marts and you’ll still see plenty of successful businesses.

In the Wal-Martification post I urged organizations to change their ways, but three years later I see history repeating itself. Look at your virtualization and private cloud initiatives. I frequently find myself as a minority voice holding the position that heterogeneous virtualization is a bad idea for production server workloads. In discussions with clients I often raise the following issues:

  • DR complexity (capacity management is more complex when you need to ensure available capacity at each site on a per-hypervisor basis)
  • Reconfiguring/replacing operational software
  • VM conversion, driver replacement and scheduled downtime
  • Vendor support
  • Organizational processes and governance, and the creation of new management silos (many private cloud initiatives result in the collapse of management silos, while it’s possible that heterogeneity can create new ones)
  • Quality assurance – oftentimes each hypervisor stack requires a new QA check due to the differences in performance validation, configuration, and operational management requirements

Multi-hypervisor is an effective strategic direction to go from one virtualization platform to another, but it has serious tradeoffs if it’s the end goal for the production server workloads in your data center. Additional hypervisors for one-off siloed initiatives is often practical, but becoming less standardized in your data centers is anything but efficient.

As you further build out your private clouds, will you follow the service providers who seem to have the SADD antidote and go with a highly standardized infrastructure stack? Or will you go the heterogeneous route? There’s a huge community hoping that you make your private cloud as non-standardized and complex as possible. Their profits depend on it. What are you going to do? Am I out of my mind on this one?


Comments are closed.