IO’s leadership recognized the magnitude of the challenge but believed that the time was ideal to attempt such a transformation. Advances in nascent technologies, especially cloud computing and mobile connectivity, were making traditional IT-infrastructure delivery mechanisms obsolete and permitting new levels of efficiency. By adopting a single internal cloud technology, for example, and forcing its application teams to build to this environment, the company believed it could move more quickly to a standardized infrastructure and make better use of existing capacity. Embracing the cloud more aggressively would also narrow the set of skills that IT needed to maintain in-house, especially in engineering and design, and would reinforce the company’s move toward a more standardized and streamlined environment. Making greater use of mobile technologies, especially tablet-based ones, would also be advantageous, the company believed, because the more standardized architecture that these devices employ would allow IO to significantly reduce the number of versions it needed to support.
The leaders of the IO function also believed that new entrants into the “technology solutions” supplier market, including Amazon, Google, and Rackspace, were introducing creative solutions. And it noted that increasing competition among these players was driving down prices across the board. All told, the time was right to move. By leveraging these technologies and making greater use of external suppliers, IO would be able to aggressively streamline and standardize its hosting environment. And it could do so while maintaining or even improving IO’s performance and cutting costs materially.
Armed with that confidence, IO’s leadership took to the task of simplification. It appointed a team of architects to examine the 1,700-plus technology patterns currently in place and identify commonalities and opportunities for standardization. (See Exhibit 2.) The team first grouped the patterns into roughly 120 metapatterns based on similar high-level characteristics. It then identified the top 20 such metapatterns, which it determined would cover the vast majority of existing applications; identified the specific standards that would need to be defined to cover most applications; and worked with subject-matter experts to define full technology patterns for each standard.
Ultimately, the team determined that a very small number of standard technology patterns—seven—would be sufficient to cover approximately 80 percent of the company’s application needs. All seven would make extensive use of shared assets, including databases, middleware, Web servers, and Web services—a sharp contrast to the original 1,700 patterns, 90 percent of which were independent and which often ran on their own hardware or software stack and were not shared. The architects also defined 13 “special use” patterns to cover exceptions (for example, mission-critical and performance-intensive environments, such as the SAP environment and plant-operating applications) that would require dedicated resources.
The move to an aggressively streamlined, standardized, and virtualized hosting environment will, IO’s leadership believes, deliver a range of outsized benefits. (See Exhibit 3.) For example, moving to an internal cloud will increase server utilization, which will greatly reduce the server footprint and associated managed-service-provider costs. Limiting the number of technologies will reduce the time it takes to design new solutions; it will also permit the sharing of resources and simplify learning curves. The use of preprovisioned configurations that have been engineered for use in many situations, as well as a fully automated provisioning process, will shorten cycle times. Ultimately, IO’s leadership expects the effort to deliver cost savings in the core data centers and remote sites of roughly 15 percent over the next three years.
The effort is well under way. Teams are working to identify and execute quick wins. Suppliers are working to establish the necessary environment. And some applications are under reconstruction. Pending next steps for IO’s leadership include applying standard patterns to all new applications and refreshes and implementing a governance process to halt any nonstandard development early in the process; communicating standards internally to facilitate rapid adoption; creating a detailed business case and timeline for migrating existing applications to the new paradigm; and developing the necessary pattern life-cycle management and governance procedures.