Forrester and Maintenance Cost Reductions

An interesting research piece on the 'application modernization' space was published this month by Phil Murphy at Forrester. He discusses approaches that development organizations can take to in order to slash application maintenance costs. You are likely familiar with statistics that show 70% and more of budgets being devoted to ‘lights-on’ activities. Clearly, savings in the application maintenance world can have a big impact by redirecting resources toward higher-value activities, like application modernization. And also by demonstrating IT’s commitment to cost savings.

This particular piece takes an insightful and inclusive approach, referencing many previous research pieces for APM and application modernization. On that point alone, this is a worthwhile reference paper. But in addition, it has other aspects that are of interest. In particular, the document looks at how dependency mapping can help.

As applications have grown increasingly sophisticated the interrelationships between various artifacts has become more complex. One program may depend on inputs from dozens of other artifacts, and its outputs may have downstream consequences for dozens of other artifacts. So, enhancements or modernizations to one program can have often unintended repercussions on the rest of the application portfolio. This can disrupt business processes and can lengthen the time required to make enhancements as analysts must research their application portfolio in exhausting detail – often without documentation or subject matter expertise.

The paper makes an important suggestion: use dependency mapping technology to uncover interrelationships between artifacts. This allows users to quickly trace all relationships, provide a list of potential impacts, and correct these impacts before they have negative consequences. While the paper recommends run-time mapping, it is important to consider static analysis. Because static analysis relies on tracing all logical paths from a target element, you can be certain that even infrequently used paths are recognized as dependencies. Run-time analysis will only spot less commonly run (but often more important) processes (like quarterly finance roll-ups) as they are run – which may not be often.

Dependency mapping using static analysis also has other benefits, notably from an architectural perspective. First, let’s say we are interested in simplifying our architecture due to the complexity that stems from our globally distributed development organization. We begin by abstracting our software based on how it is managed. So, artifacts that are managed by the same team are grouped together. This abstraction layer provides a summary of the various interdependencies.

We can then immediately see that there are too many dependencies between the software managed by India and Ireland. From a practical perspective, this means that each time India makes a change it may need to contact Ireland to avoid unexpected impacts. This could lead us to rearchitect / modernize our application portfolio or reallocate assets so that we can minimize these cross-geography dependencies.

Gartner on APM Best Practices

An excellent piece of research came out from Gartner this week. Jim Duggan’s research details methods to make Application Portfolio Management successful. A key aspect of the research centers around the need to manage from diverse perspectives. That is, to manage the application portfolio based on contexts that matter to the business. A post on our Application Portfolio Management sister site goes into greater detail.

Let’s outline why Application Portfolio Management is so important. IT has no shortage of priorities. But are they the right priorities? Are investments being made in the right way to support the needs of the business? It’s hard to select and justify projects unless you can validate that they support corporate priorities. And with the trend toward cost containment this is becoming a more important issue. In the same line, IT must be investigating where cost savings can be achieved by rationalizing systems that can be eliminated.

That is where Application Portfolio Management comes in. It is a process of measuring the value, cost, and risk of the application portfolio from perspectives that are important to the business. For instance, what is the cost of applications that automate a given business process – versus its value to the organization. Or, what is the software quality trend for our outsourcing partners.

There are two keys here. First, we are collecting metrics from diverse sources. It could be stakeholder surveys, code metrics (like function points or cyclomatic complexity), or cost data from an ERP system. The point is to collect the kind of measurements that will aid the decision making process. Second, we want to place our collected metrics into the right business context. That is, we want to collect and group the metrics such that they are organized by business concept. This allows us to make IT decisions in the context of the business goals that we are supposed to manage to.