Application Discovery – Industry Viewpoint
Application discovery, required as an initial input to data centre migration, cloud migration, and application consolidation/rationalization programs, is usually a ground-up green field activity. Even the most organized businesses have only a partial inventory of applications, and most often this information is inaccurate and dated. Coupled with incomplete views of infrastructure (server inventory, application mappings, etc.), we are running blind.
Enterprise migrations to either cloud or data centre environments, requires a combination of top-down application discovery, and bottoms up infrastructure discovery. Application discovery in an enterprise computing environment is generally a last resort effort to understand the environment. Usually after extensive use of discovery tools and manual methods have been exhausted (usually with underwhelming and costly results), do customers accept the time and cost associated with top-down discovery.
So, to close the gap and inventory an enterprise application ecosystem, let’s take a quick look at some of the root-causes behind our poorly rationalized current state:
Organization / Cultural Divides
- Application and infrastructure teams most often are divided, with lines of reporting only converging at the CIO
- With this, you also have a massive cultural divide where applications teams have little interest in infrastructure, and infrastructure teams generally having little to no visibility into the services/applications running on their kit
- Application documentation is most often non-existent, outside of support procedures
- Infrastructure documentation typically originates from design/build projects, but is rarely maintained
- The classic CMBD is most often the assumed repository for applications, application ownership, and infrastructure item mapping
- The CMBD is usually the ‘trusted source’ but in reality poorly maintained and at best 50% accurate
- Application ownership, even if once documented, changes over time and attrition. Couple this with turnover, outsourcing, offshoring, and the end-result is little to no knowledge of the application fleet in production (at least from an infrastructure point of view).
- Another variable is the evolving nature of application environments. Over time, new functionality is introduced, new application components, 3rd party tools, all spawning modified versions of the original build. Who’s keeping track?
The solution is a combination of manual discovery for applications, and tools based discovery for infrastructure. No one tool or automated procedure truly identifies an application, maps interdependencies, and infrastructure alignment, while producing actionable information for business or technical decisions. So we deploy a combination of organic and automated methods for a complete discovery of enterprise applications. A quick summary as follows:
- Systematic interviewing of application group owners
- Consistent documentation and metadata management of results
- Validation of infrastructure mappings & interdependencies by application owners
- Application owner signoff on application stack and business requirements
- Physical site auditing and inventory
- Tool-led discovery of physical and virtual infrastructure environments
- Interdependency mapping (for validation by application owners)
- Common repository for infrastructure and application metadata
For an enterprise, detailed discovery and validation may take many months, depending on budget, complexity, and organizational maturity. While time consuming and costly, base discovery is the immediate driver for consolidation/rationalization and subsequent migration to modernized compute environments.
For more information or to discuss this further, feel free to engage.