Peter Gallagher | System re-use'an LOB we can all agree on
- By Peter Gallagher
- Apr 24, 2008
Progressive Disclosure columnist
The vision for immense savings by agencies re-using software has been something of a holy grail in government. Yet despite significant investments over many years in object-oriented and modular programming, repositories like www.Core.gov, and more recently, service-oriented architecture (SOA), federal, state and local governments have largely failed to generate those savings.
Many federal agencies, including the Defense Department, have invested in efforts to share software and related system assets, like requirements definition and other documentation that are germane to their domain. The Office of Management and Budget has tried to encourage software re-use with its push for Federal Enterprise Architecture and the potential to identify common service components. The ambitious effort behind the Lines of Business initiatives, also lead by OMB, was designed in part to produce re-use savings through shared systems and services. The General Services Administration's Smartbuy program also attempted to tackle this re-use challenge indirectly by consolidating governmental purchasing power.
Why is something that seems so logical - re-using software - so difficult in practice? A big reason is the complexity of the task: It is impossible to coordinate and control the creation of software applications governmentwide and any effort to do so centrally is certainly doomed. That's one reason that there is no central policy mandate to require or explicitly value the re-use potential of information technology investments.
Additionally, the proprietary interests of vendors and systems integrators tend to keep the software assets under wraps to enhance competitive advantage. A proprietary approach makes economic sense for the companies that want to generate maximum profit by re-selling the same software. But it naturally clashes with the government's desire to minimize expenses and allow the government to "buy once and use many."
System and software re-use is most frequently achieved via market forces building on commercial off the shelf products. Of course these COTS products often require extensive integration and/or customization, creating new software code, but current policy has long favored COTS over custom-built or government/GOTS for good reason. Today, the vast majority of the COTS used by government are based on the proprietary model where the software asset is closely held by a sole licensor or vendor. Often, even highly customized software created specifically for government purposes is proprietary or has re-use restrictions. And even when the software is delivered with full Federal Acquisition Regulations "rights in data" the government cannot easily share what it owns in large part because there is no mechanism for competing vendors to leverage.
Commercial support for re-usable systems and software, now potentially broken out into smaller component services via SOA, requires a model where continuous improvement can take place via dispersed resources. Orchestrating such a governmentwide process would be an overwhelmingly complex management task. However, the expanding market for commercial open-source software (OSS) has been able to make sense of just such complexity. Commercial OSS products now dominate some of the most mission-critical software functions. We need look no further than the Apache Web server for proof that shared resources can be leveraged to produce leading-edge products. Apache is in fact the standard at some agencies. And increasingly in the private sector this shared-resource model is being applied to maximize profits. Google's recent effort to build a new platform for cell phones is a good example but there are many others in progress.
What if such an open commercial model for software asset sharing was applied regularly to areas with strong government data standard stewardship such as information sharing tools for health records, justice information, employee/contractor identification or learning content? Take the essential Domain Name System protocols that connect the Internet via the BIND software - BIND is open source and was in fact backed by the Defense Advanced Research Projects Agency and the Defense Information Systems Agency as well as commercial interests. Couldn't the commercial open-source model be applied to reduce these costs and make government systems better, faster, cheaper? Why do government agencies - federal, state and local - keep buying the same systems over and over?
Coordinated attention by the feds in promoting re-usable software using commercial models has the potential to show huge savings. That is a Line of Business that we taxpayers could all agree to support.
Peter Gallagher, now a partner at Unisys in the Federal Civilian group, has been working to promote the use of open source software and open standards for government since the late 1990s.