By Harris Kern

Businesses have enough competition outside the business
without having it inside as well. I propose creating an internal service level
agreement called an internal support agreement to keep your infrastructure
support group and your applications development staff away from becoming

In many of the companies we visit, the applications
development staff is located at the division or business-unit level. Unfortunately,
the support groups reside in different buildings. How does a centralized IT
infrastructure support group provide System Administration services to
scattered corporate developers? This and other issues should be clearly defined
in an internal support agreement.

Tips in your inbox

TechRepublic’s free CIO Issues newsletter, delivered each Monday and Wednesday, covers C-level managerial advice, from HR best practices to infrastructure management tips.

Automatically sign up today!

Of all the battles inside IT, none have been uglier than
those between applications development and the infrastructure support
organization. Even when development and support were centralized under one
organization, there was always finger-pointing. The charter for applications
development is to design, develop, and deploy systems into production as
quickly as possible. The charter for infrastructure support is to ensure that
proper standards, guidelines, and QA testing are adhered to, and that thorough
documentation is provided. Most of the issues arose around implementation and
support of mission-critical applications. Development would blame Operations
for messing up a restart to a system error, or Operations would blame
Development for lack of QA or support. There were many other issues of this

As in the mainframe era, developers still want the
centralized IT infrastructure staff to support development servers in a limited
way for such mundane tasks as backup-and-restore. But today’s development
servers are a problem for both the development group and infrastructure
support. Developers want top-level, “root” or full administrative
access to their development machines. This is a problem because owners of root
passwords can bypass normal safeguards and unwittingly destabilize a machine in

On the other hand, developers do need occasional unlimited
access to a machine. Denying access makes their jobs more difficult than
necessary or even impossible. What to do?

To solve this problem, we devised joint root authority. The infrastructure support organization and
the two most senior developers will own root access. If developers abuse root
privileges, the infrastructure organization will no longer support development

One of the most important pieces of this puzzle is an internal Service Level Agreement (SLA)
between the infrastructure support staff and the applications development
staff. In this document, expectations are clearly outlined for both groups. The
following sections show some of the key categories and examples of an internal
SLA. When designing an Internal SLA (for supporting a development server) the
first area to start with is defining security privileges.

SLA elements to define

Server root authority

Root access to support servers AD0001 and AD0002 in the
following example is given to Mr. A and Ms. B. Mr. A and Ms. B are to
support/back up each other in case of illness or vacation. If they’re both
unavailable, you would contact Technical Services.

All changes to a server’s root will be audited to provide a
trace of root user activity. The following activities are accomplished by
Technical Services upon request:

  • Kernel changes
  • Disk reconfigurations
  • Modifications to the root user environment
  • Installation of binaries into the system
    directory structure
  • Modification to any network-related
    configuration files
  • Manipulation/modification of any system daemon
    run at the root level
  • Changes to the /etc/rc* files

The following is a partial list of activities that may be
accomplished by the applications development root owners:

  • Change /etc/exports for mount directories
  • Change /etc/fstab
  • Add users/groups