The Cloud team shaped a cloud strategy and plan, which was to adopt, optimize, and transform. The focus of the first phase was to accelerate cloud adoption through 1) rapid migration of business applications out of on-premise data centers and into the cloud and 2) diligently adhering to a “cloud-first, cloud-only” strategy and principle for all new applications. All new internal IT capabilities were required to build their business applications in the cloud, unless they requested and were granted an exception. The internal IT organization did not grant any exceptions during this journey.
The first—and aggressive—target was to move from 9 to 50 percent operating in the cloud within one year. The team took the impactful approach of migrating Accenture’s highest-cost data centers in North America first, even though it represented a greater risk due to the migrations involving production environments supporting business-critical applications. The first major public platform to go into the public cloud was Accenture’s own website, Accenture.com.
For the first data center migration, the team had to overcome the challenge of not having a full solution for migrating complex enterprise applications at the desired pace. To address this, the team partnered with the Accenture Cloud Factory to test and execute new technical migration procedures using the Accenture Cloud Platform (ACP) Accelerate framework. As part of ACP’s suite of pre-integrated tools and services, Accelerate consolidates best practices and lessons learned for cloud migrations.
The team also created, standardized, and trained people on the supporting processes required to execute successfully in a highly matrixed and varied business stakeholder and application environment. What resulted was a fully tested cloud migration solution with new methodology addressing these technical and non-technical enterprise complexities that is today being offered to clients. It was used for the remaining internal IT on-premise data center migrations.
For the second data center migration, the team partnered with Accenture’s Internal IT Disaster Recovery team to migrate from a traditional capital-intensive co-location recovery data center to a cloud-based data center where recoveries can occur using a lower-cost solution. The migration was completed in eight months, and resulted in greatly reduced infrastructure cost, improved manageability, and improved recovery point objective (RPO).
Throughout the third, fourth, and fifth data center migrations, completed in 2017 and 2018, the team continued to refine processes and tools, ultimately leveraging a mature solution. Along with ongoing business application migrations, the second rung of the cloud adoption strategy, “cloud-first, cloud-only,” continued to drive cloud adoption in parallel. The strategy required all new applications to be architected and designed for the cloud, and more than 95 percent of all new infrastructure is provisioned directly in the cloud.
The other lesson the Cloud team learned quickly was a “less is more” in the public cloud. This approach focused on taking IaaS workloads and instead of migrating like for like, the team did like for less. The cloud at that point was maturing quickly and Accenture was able to immediately get savings even doing simple lift-and-shift activity.
With the technical foundation set and cloud adoption well under way, internal IT’s focus shifted to optimizing what Accenture had adopted. Optimization opportunities were first pursued in infrastructure hosting and tool automation and progressed to further focus on process streamlining and governance simplification.
A key difference and opportunity in the cloud hosting model is that resources only need to pay for what they consume. Most cloud migrations get this wrong and therefore get in trouble quickly with the spend. The flexibility of cloud hosting allows applications to request more or less on demand. This means that instead of overestimating to create large contingencies, teams are enabled to react and adjust to real-life circumstances and to proactively and aggressively avoid unnecessary hosting costs.
For example, nonproduction cloud server schedules are standardized to reduce unused “uptime.” Understanding that every application and customer has different circumstances, teams are enabled to start and stop their nonproduction cloud machines on their own, without having to go through manual tickets and procedures—a step to self-service transformation. Similarly, teams have influence to actively manage the actual server size of their cloud machines. This is a major lever in controlling cost, so internal IT made it a mandatory step in our provisioning cycle. A team requesting compute had to, as part of that request, outline the actual up/down schedule that would be automated.
The cloud also offers creative solutions for hosting pricing and allocation. The Cloud team took advantage of different optimal hosting pricing models for scenarios where demand was known and consistent, as well as for demands that had a pattern of short-term spikes.
Finally, the team even identified opportunities to save by moving to dedicated cloud-hosted hardware in order to leverage existing software licenses. These opportunities enable teams to influence hosting costs and make decisions that treat efficiency as a key value driver.
Overall, continuous optimization is part of Accenture’s evolution. All of these optimizations lead to business benefits that contribute to lasting transformation.
To be better equipped for the pace of evolving cloud technology and capabilities, the Cloud team restructured into dedicated teams that adapted agile delivery methodology to deliver the transformation. This change allowed infrastructure teams to operate more like product or application teams with better stakeholder representation throughout delivery. Agile delivery is usually implemented by singular, co-located software product development teams that execute daily sync-ups, incremental development, and frequent delivery. In the Cloud team’s case, it committed to transforming toward agile principles, but needed to do so in an environment that is not traditionally agile—globally dispersed, infrastructure-focused, and with multiple technology functions and owners.
In addition to applying agile delivery principles, the team restructured into dedicated teams that focus on:
- Conducting research, proof of concept, innovation, and solution engineering activities.
- Evaluating infrastructure requirements such as network, security, and environment support.
- Developing and deploying cloud management tools according to the cloud tools road map.
- Executing a flexible but controlled process for cloud service adoption prioritized by business needs.
- Operational support once new solutions are adopted.