Blog post

More IT Centralization May Not Be The Best Route to Cost Saving

By Andrea Di Maio | October 01, 2010 | 4 Comments


As reported by Information Week, the Governor of California Arnold Schwarzenegger has signed a law to improve the state’s IT infrastructure and make it less expensive. the law aims at stronger project oversight, increased transparency in spending, greater cost savings, energy usage reduction and further services consolidation. It strengthens the former Department of Technology Services, now transformed into Office of Technology Services under the California Technology Agency.

Among the provisions of the law:

  • consolidation of data center space by 50% by July 2011
  • cut energy usage from IT operations 20% by July 2011 and 30% by July 2012.
  • standardization of IT governance under the Office of the State CIO
  • hosting all mission-critical public-facing applications and server refreshes in a Tier III data center, closing all existing data centers and server rooms that house non-network equipment by June 2013.
  • migrating from existing network services to the new California Government Network no later than July 2011, and begin migrating to a state-shared email solution no later than June 2011.
  • development and maintenance of enterprise architecture plans to comply with statewide enterprise architecture policies and standards as established by the California CIO,

At first sight, the list above makes a lot of sense. Reducing duplication, increasing compliance, better utilizing resources, further consolidating and centralizing are all important steps that need to be taken in order to reduce costs in a sustainable way. A great advocate of centralization has been the State CIO Teri Takai, who has been selected for a federal role at the Department of Defense, but may not leave as soon as expected (see NextGov latest news).

On the other hand, is centralization the only option to achieve savings, or should a better balance be found between state-wide decisions and the ability of individual agencies to find better and cheaper solutions? For instance, in areas where the market is commoditizing, wouldn’t it make sense to let agencies make their choices, and focus on controlling how they procure services rather that necessarily provide services from an internal organization? Excessive centralization can lead to tension, slower than planned deployments, and the constant pressure for the central organization to be better than anybody else. The common wisdom says that consolidating the buying power and service all agencies permits the achievement of economies of scale that would not be otherwise possible.

However, does a single data center work better than a hybrid solution where agencies have the autonomy of buying – for certain categories of applications – infrastructure as a service straight from the market?

Does the focus on enterprise architecture encourage or rather prevent different agencies from exploring innovative solutions that may slash the cost of their operations?

Isn’t there a risk that all this centralization makes agencies feel that IT is really a commodity, take a compliance attitude to meet the requirements of this law, and lose countless opportunities for transformation?

While I do understand the background that led to this decision, and I know that many of my own colleagues always applaud to more centralization, I wonder whether this is not going too far, reducing the ability of exploring those radical (and innovative) cost cutting options that a jurisdiction like California may have to face earlier rather than later.

Comments are closed


  • Bryan Sivak says:


    Your point is well taken but this can be gotten around by making sure agencies and local jurisdictions have an “escape hatch.” Here in DC we are centralized in all of these ways (email systems, eventually data centers, networks, etc.) but every agency is free to put forward a detailed business case for breaking these rules. If they can do it faster, better and/or cheaper than my agency’s centralized efforts (both initially and ongoing), I should not only allow them to do it but also learn from it and try to replicate or offer the service centrally.


  • Good to hear from you Bryan. I guess it is a matter of perspective. Irrespective of how much is already centralized, does further centralization necessarily help? In your case I would argue that the level of centralization is already higher than in California, but would you aim at centralizing even more as a means to save money? You make a great point about how central service organizations can evolve by learning from what their own users do when going on their own (of course, with a decent business case)

  • Bruce says:

    While we are not a government based, the same concept and issues applies to an enterprise with different divisions. A centralized infrastructure and subject matter IT experts does not preclude decentralized decisions or options. Let those agencies or departments seek out and look for the best solutions to their specific needs, but also allow the central authority to ensure that these outside services are technically sound, secure, and a duplicate of a service already available or provisioned somewhere else. e.g. . You don’t want an agency looking for a cheaper email system that is inconsistent with enterprise standards and will result in a Tower of Babel.

    To have a successful hybrid there needs to be mutual respect and understanding between all parties. Central IT groups tend to say \ I know what’s best for you\, but they may not. So indeed it is give and take and a learning experience for all.

  • There are tradeoffs between centralized and distributed ICT; having been in ICT for 25 years, I’ve seen the pendulum swing between the two, depending on the availability of money. Have a look at the diagram I did here,, showing advantages/disadvantages of both approaches.

    Also google Steve from Ovum, who wrote a paper called the egovernment Chimera, which outlined the particular tensions that public service CIOs face.