Chris Wolf

A member of the Gartner Blog Network

Chris Wolf
Research VP
6 years at Gartner
19 years IT industry

Chris Wolf is a Research Vice President for the Gartner for Technical Professionals research team. He covers server and client virtualization and private cloud computing. Read Full Bio

Coverage Areas:

Standardization ADD: We are all Hypocrites

by Chris Wolf  |  June 21, 2012  |  8 Comments

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?

8 Comments »

Category: Cloud Server Virtualization Virtualization     Tags: ,

8 responses so far ↓

  • 1 Tony Miller   June 21, 2012 at 2:10 pm

    I have struggled recently with associates to properly convey this ideal across to them. Thank you for effectively describing the conundrum for others.

    It boils down to the point of diminishing returns (or savings?). What is that 20%-33% savings up front really costing? I can explain all the values of standardization and hardware life-cycles until I am blue in the face, but its difficult to show the true ROI of the investment to the decision makers because it happens over time. The reduced downtime, ease of management, interchangeability and support will have significant savings in man hours. Man hours that could potentially allow for progression forward as opposed to “reinventing the wheel”.

  • 2 T Benson   June 21, 2012 at 4:51 pm

    Hi Chris,

    I thoroughly enjoyed your recent talk at the VMware conference in DC! It was a nice wake up and listen moment.

    In all discussions, a serious thoughtful overview is highly valuable. Thank you.

    This article is very good, but it’s also a little hard to process, if you don’t mind my saying so. I think you could tell this story in a lot of ways. Your talk this week really got the message across quite eloquently!

    Thanks again for your excellent talk and provocative truthfulness–it helps all of us to make more logical decisions.
    T Benson
    tbenson314@gmail.com

  • 3 Cesar Torres   June 22, 2012 at 4:52 pm

    Chris,

    This is great, I agree with you 100%. It was nice talking to you the other day.

    Take care,
    Cesar

  • 4 Doug Baer   June 22, 2012 at 7:30 pm

    Chris,

    I totally agree with you, and generalize it even further than hypervisors:

    “Additional for one-off siloed initiatives is often practical, but becoming less standardized in your data centers is anything but efficient.”

    There is a nasty misconception that standardization is bad because it equates to vendor lock-in. Because we have the capability to migrate between hypervisors fairly cleanly (it helps a lot to know what the source and target ‘hardware’ looks like), there is a valid exit strategy in the event that any given vendor starts to get too comfortable. I don’t think this means that anyone should increase complexity for complexity’s sake: pick a platform and use it.

  • 5 Chris Wolf   June 22, 2012 at 10:09 pm

    @Cesar, Tony, and Time – thanks for the feedback. I appreciate it.

    @Doug – I agree with your point that hypervisors is one example. I think the problem is pretty pervasive across several aspects of IT.

  • 6 Massimo Re Ferre'   June 26, 2012 at 11:42 am

    Amen.

  • 7 Dave Convery   June 26, 2012 at 1:34 pm

    Chris –
    I must add to Massimo’s AMEN! I am constantly asked to compare different vendors and which is “better.” Or I am asked how to make things work that really shouldn’t work. When I ask about why someone wants to do something, most of the time there is an answer that turns out to be an “I want” rather than “I need.”

    There is a reason why systems like the AS400 and Mainframes still exist today. Standardization is good for the business of running IT. In the X86/x64 world, it seems like there is always some sort of turmoil about a new technology. Its great for people like me, because this causes unplanned “drag” of more stuff to sell. We have to kludge together a buch of stuff to make your heterogeneous environment play nice. While your IT staff are busy hammering square pegs into round holes and worrying about managing hypervisors, your “customers” (developers, users, etc.) are renting stuff from Amazon because it is easier than dealing with IT.

  • 8 Dave Convery   June 26, 2012 at 1:34 pm

    Chris –
    I must add to Massimo’s AMEN! I am constantly asked to compare different vendors and which is \better.\ Or I am asked how to make things work that really shouldn’t work. When I ask about why someone wants to do something, most of the time there is an answer that turns out to be an \I want\ rather than \I need.\

    There is a reason why systems like the AS400 and Mainframes still exist today. Standardization is good for the business of running IT. In the X86/x64 world, it seems like there is always some sort of turmoil about a new technology. Its great for people like me, because this causes unplanned \drag\ of more stuff to sell. We have to kludge together a buch of stuff to make your heterogeneous environment play nice. While your IT staff are busy hammering square pegs into round holes and worrying about managing hypervisors, your \customers\ (developers, users, etc.) are renting stuff from Amazon because it is easier than dealing with IT.