Focusing on IT infrastructure instead of focusing on ways to help your company grow is one way to keep IT being viewed as a cost center.  Here are some tips to break that cycle.


In IT, a large portion of our lives are spent dealing with infrastructure issues.  Infrastructure is a tactical, keep-the-lights-on type of thing that keeps you as the IT leader from focusing on ways to help your company grow.  The more time you spend on infrastructure, the less time you have applying technology to business problems and opportunities and contributing to revenue-generating activities.  This is likely why IT is seen as a cost center in most organizations.  There’s a way to get strategic about infrastructure, but the first step is to minimize how much of your day it sucks up. The steps:

1. Standardize! Desktop operating systems, office productivity tools, software, servers and network equipment. Standardization has several benefits:

  • Usually, you get better pricing if you go through one partner.  As a small company, you’re not going to get huge volume pricing discounts from the Dells of the world, but many of these companies have VARs that really give great customer service.
  • The systems have been tested to work together. Guarantees of interoperability are hogwash. Servers have kernel level utilities that may make an HP server perform differently than a Dell.  Having a mixed farm for your website or ERP system is a level of complexity you don’t need.
  • Supportability. I’ve found it’s better to have employees with deeper knowledge of fewer systems and applications than to have a bunch of IT generalists. An expert in Cisco IOS can find problems faster than a network generalist who may need to open a manual for an HP ProCurve switch and a Cisco router. This is especially important for core systems. At one job, six weeks after I started, there was a power outage that crashed the e-mail server. One of our system admins had the manual open for Exchange and it took three days to get it up and running and then it had two weeks of other minor issues. I had a Microsoft expert do a reinstall, an upgrade and cluster on that same server and it was done in four hours with no issues. Cheap is not necessarily less expensive.  Obviously, the IT specialist scenario does not work if you have hundreds of applications.  Reduce the number of systems to as few as possible and get specialists to manage and maintain.  It simplifies everything.
  • Less administrative work. In an SMB, you’ll likely not have an administrative assistant. That means contracts, maintenance agreements, etc., all have to be managed by YOU. Make less work for yourself. Standardize with partners and not vendors (future Blog entry on Vendor vs. Partner) and limit the administrative headaches.

2. Eliminate variability. Variation to any system is the enemy and is at the root of 99% of bugs, crashes, and system hiccups.

  • Create your configurations (switches, router configurations, server builds) and put them in source control.  This ensures that in the event of an issue, there is always a steady state to fall back to.  An added benefit is using the common tools within the source control system to compare configurations and identify changes.  Versioning of configurations is a huge benefit as well.
  • Have every system with a control book and follow them. Any change to a system or a server will have some unanticipated impact, but not knowing what change has been made to what server is an entirely unnecessary headache you don’t need.  In an SMB, control books are a pain in the neck.  The number and scope of changes to a system can make this documentation process overly time-consuming, but it will sure save you when something goes wrong.  Notice I said “when” and not “if.”  If you’re making that many changes to a system that documenting becomes time-consuming, then you need control books more than anyone.
  • Make sure all changes go through a change management process. No change should hit production unless it has been thoroughly tested and the control books updated. Have a backup plan for the backup plan. Murphy’s Law applies here. If something can go wrong it usually will.  Virtualization is a great and inexpensive way to build, test, and deploy changes to a production environment.

Guard against variation and fight hard against any pressure to introduce it to a stable environment. This may be one of the hardest things you have to do in your job. Have the discipline to say no. You’ll live a much happier life.

Following these two rules of thumb will solve 80% of your issues because they are typically the root cause to the bulk of your headaches.  Now you can focus more on helping your company make money instead of perpetuating ways for your company to spend it.  My next post will talk about creating an infrastructure that can be strategic.