Data Centers

CIOs must get their infrastructure in order

Though time seems to be in shorter supply than ever before, it's imperative that CIOs focus on their IT infrastructures if they expect to be successful.


By Harris Kern

Due to global competition, business requirements/issues, shrinking IT budgets, and unhappy customers, CIOs have no choice but to focus on getting their infrastructure in order. Let's examine the main obstacles to the effort to make operations cost-effective and methods for overcoming them.

Obstacle 1: Lack of long-term planning
Infrastructures are bending, bleeding, and becoming more costly than ever because everyone’s supporting more disparate systems than ever before. We're spending very little time on implementing key processes. Actually, the organization structure is so far off center that processes would not be effective in this environment anyway.

CIOs everywhere have been focusing on new systems development and deployment because they know that development is where they get the biggest bang for the buck. Also, the quicker new systems are deployed to meet the needs of the business, the easier it is for the CIO to gain peer respect in the short run. Unfortunately, this modus operandi doesn't encourage long-term planning.

What about the long-term effects of continuously adding network information systems? If the new applications don’t fit into the overall IT architecture or don’t have ample bandwidth or adequate support from the DBAs, they don’t work. Believe me, I’ve seen too many examples of well-intended projects going amok because they were created in vacuums without consideration of the infrastructure required to support them.

Obstacle 2: Executives out of touch with their infrastructure
Things are desperately out of control with the IT infrastructure. The problem is that executive management is overloaded, but now it’s become a crisis.

As a first step in fixing this, CIOs need to dedicate a few days to understanding the issues their organizations are facing. Spend some time in the trenches with your technical staff, not management, to get the true picture. It will be a wakeup call that you’ll never forget.

Establish metrics
As a CIO, you have to establish a set of common metrics to judge how well the organization is meeting its daily, short-term, and long-term objectives. With common metrics, everyone in the organization will speak the same lingo. Think about the data centers of the 1960s and 1970s. Typically, the data center was one location where everyone worked together.

Today, with applications development staffs scattered around the world, the data center has become a compendium of different data centers, and executives are not located together. It’s no wonder that people are out of touch with one another—and that executives are so easily out of touch with their organizations. The point is for everyone to be in touch with what’s happening and to have up-to-the-minute news and measurements for performance so that management can be proactive, not reactive.

With a common set of metrics, everyone on the IT team can immediately gauge how the team is doing against its own measurements. My contention is that you achieve high-level reliability, availability, and serviceability (RAS) only when you understand what the measurements mean. Too often, I’ve seen CIOs gloss over the nuts and bolts of making RAS possible. I’ve seen executives assume that RAS is the responsibility of lower level data center and network managers without understanding the linkage of the different components that make this possible. With a common metric system for RAS, managers can see how different departments are performing and they can correlate results among departments.

Ensure a RAS environment
Supporting client/server applications encompasses multiple hardware and software components that must be monitored to ensure a RAS environment. Products such as Hewlett Packard’s OpenView and Computer Associates’ UniCenter can provide seamless interfaces to hardware and software products to gather operational statistics and other data. Products like Microsoft’s MSM can provide for desktop management and ensure that only approved software is loaded on desktops.

Information about the network-computing environment is critical for RAS. You should include in your dashboard of information basic data points such as:
  • Assets inventory (network-wide hardware and software)
  • Network connection(s)
  • Network performance (including data packet management)
  • Desktop software (inventory control)
  • Database(s) connections
  • Database(s) availability (session status)
  • Database(s) performance
  • Operating system(s) availability
  • Operating system(s) performance

Obstacle 3: Lack of time and resources
There is a severe lack of time and resources for building the proper infrastructure. New technology or new systems continue to take precedence. However, the proportion of 90 percent applications development and 10 percent infrastructure development will be the death of network computing because the systems will continue to bleed and eventually fail, and so will their leaders.

CIOs I’ve worked with claim that because of their organizations' large staffs and budgets for infrastructure, they spend nearly half their time on the infrastructure. But, they also admit that they focus most of their energies on applications development, not really understanding the issues with their infrastructure. It’s a matter of focus. Until CIOs concentrate on improving the structure of their organization, they won’t have RAS with network computing.

The role of the CIO is to harness the available resources, decide what projects can be undertaken to succeed, and set up a viable system to measure performance on a daily, midrange, and long-range basis.

Assure that your IT budget provides for RAS
As the leader of an IT organization, it’s up to you to make sure that the budget has ample production support services to attain RAS. Many application promotions become disasters because they’re created without regard for budget costs and the additional overhead for production services.

Many costs are incremental and are, therefore, not full costs. But you must budget them as full costs. For example, the addition of an application to the production portfolio may add only a workload equal to .5 full-time equivalent (FTE) persons in the database administration support group. Unless the organization has an arrangement with a part-time DBA service provider, however, the cost of support will grow to one FTE or the existing staff will be faced with the added burden without increased personnel. In some cases, this may be handled with a budget cycle with overtime dollars, but I don’t recommend it.

The CIO must oversee the budget cycle or delegate this responsibility to someone in the organization. Either way, make sure that the budget checklist includes:
  • Operational process design, implementation, and ongoing support.
  • Implementing a production control function for infrastructure QA to preserve RAS.
  • Training costs, especially flexibility in calendar assignments so that staff can attend classes.

Marketing and selling IT services
The marketing and selling of IT services needs to be a full-time job function and part of each and every manager’s responsibility. Not having the time can no longer be an excuse. You have to take responsibility to market IT’s services and mission inside the organization.

IT still has a rotten image and network computing has made things worse. “IT is spending millions and what are we getting in return?” is the most common theme. Now, we know better—customers are getting plenty of service but they still complain and say they receive very little attention from IT. We need to get rid of this perception once and for all.

Let’s face it—in the era of outsourcing, any IT organization is in competition with other providers that might provide the service better, cheaper, and more efficiently. You need to start by thoroughly documenting services along with associated costs for those services. Walk with the great unwashed and communicate with customers. Schmooze, sell, and otherwise promote your services. Table A illustrates ways to do this.
Table A
Activity Description
Publish newsletters. Publish a quarterly newsletter to be distributed to your primary customers in the organization outlining your recent successes, updates to your business plan, and plans for the near- and long-term future. Include statistics showing improvements in supporting customers, such as:
• Uptime availability of the network
• Uptime of applications
• Savings business units enjoy as a result of applications that have been deployed to support in-line business units.
• Number of help-desk queries handled in a fiscal quarter.

Assign IT representatives to be liaisons with business unit executives, and coordinate regularly established meetings with the business units.
The CIO and representatives to business units meet with the internal customers to review their business relationships.
Arrange tours of the Data Center. Host periodic tours to acquaint your customers with your facilities and the staff.
Implement an internal IT homepage. Publish information about your IT organization, its mission, services, organizational structure, and accomplishments.
Internal marketing examples

Today’s infrastructures are more complex than ever. It’s up to the CIOs to make sure they’re cost-effective and that all executives in the organization understand IT’s mission.

For more information on the Harris Kern Enterprise Computing Institute, visit http://www.harriskern.com/.

Editor's Picks