And you would be right. But only partially.
There are a boatload of costs associated with managing far-flung IT resources, but the biggest cost-generater is complexity, said Dennis Gaughan, a research director who leads AMR’s research around IT governance issues.
“One of the benefits of centralization is you’re lowering complexity and any time you lower complexity you tend to have lower cost,” he said.
By centralizing as many IT assets (both hardware and software) as possible in as few locations as possible, complexity decreases significantly, and this leads to lower costs, agreed Scott Holland, a senior business advisor with the Hackett Group.
“Is it better to have less applications to support?” asked Holland. “Sure. Is it better to have less suppliers? Sure. Is it better to have less data centers? Absolutely, from a cost perspective.”
Distributed IT means more support and maintenance people to field and pay for, more asset and license management headaches, less control over IT spend, and reduced bargaining power with vendors, to name a few of the drawbacks.
In the ’80s and up into the late ’90s, there really wasn’t a choice: either you sacrificed speed and application availability or you put compute resources physically close to users.
Today, that is no longer the case. Broadband, remote monitoring and management tools, net-native and hosted software, and other advancements means applications (aside from desktop productivity tools) can be separated by continents from their users and still perform well, said Accenture’s Rich Melnicoff, managing partner in its North American Strategic IT Effectiveness practice.
“So all those factors coming together: the rack and stack hardware; the separation and consolidation of memory; the OS (operating system) sophistication on fail over; and the new management tools and techniques; and the physical security. Those are all driving towards co-location of hardware. And with low network costs, there is much less need to have the hardware co-located with the user,” he said.
The Softer Side
Then, of course, there are some softer costs that have nothing to do with technology.
High among them is compliance. With the Sarbanes-Oxley Act (SOX) comes a whole host of audit and visibility issues that IT has never had to deal with. And, simply put, it costs more to audit a distributed IT environment than one that is more centralized.
“You’re adding cost, your adding risk and you’re potentially adding latency to the (compliance) process as well,” said AMR’s Gaughan.
IT governance and control is also more difficult and expensive if your business units are still controlling their own IT environments, for example, or, worse, have their own shadow IT department set up to run customized applications they feel they can’t live without.
Loss of bargaining power is also an issue if five divisions are negotiating with the same software or hardware vendor for, essentially, the same products. A centralized procurement strategy can save a lot of dollars when it comes time to put pen to paper, said Holland.
“If I can do all my procurement centrally, in one area, I have much greater control of that, first of all,? he said. ?But, secondly, I have better negotiating powers instead of saying everybody’s on their own and buying their own PCs.”
Rationalization and standardization of infrastructure and applications is also made much more difficult in a distributed environment. Dictating from on high what applications can and cannot be used or purchased, for example, is fine in theory, but how do you really know that these edicts are being followed?
With most enterprises today looking to better align IT with their businesses, key to that effort is having everyone using the same software tools (ERP, CRM, SCM, etc.) in the same way. Sixty-five instances of Oracle ERP does not lend itself to a commonality of experience, but one or two does.
Losing Legacy
A less obvious side benefit of an application rationalization effort can also be end-of-lifing older legacy applications that, over the years, have been bolted onto enterprise applications, said Accenture’s Melincoff.
“Where you save a lot more money, or at least as much money, is end-of-lifing other applications which are around for just sort of historical purposes, particularly custom code,” he said.
Or try implementing an SOA (service-oriented architecture), for example, in a distributed environment. It will be difficult at best. How about a BPM initiative? Project management?
All in all, it appears clear that the longer you wait to begin the process of reining in distributed IT resources, the greater the cost. But with the tools available today to undertake this effort, it shouldn’t take much to convince your c-level colleagues that now is a good time to start.
“The key is really just defining the process and organization around how to make (centralization) decisions,” said Gaughan. “And that’s what we see a lot of companies doing … I kind of describe it as instead of having the pendulum swing back and forth, you’re trying to stop the pendulum in the right spot for you organization, at least for today.”