In early 2019, a business decision was made at an American professional sports league to migrate its content management system (CMS) and e-commerce platform to Google Cloud Platform (GCP). The professional sports league made the move because its older system was becoming more and more difficult to maintain. It also wanted to offer its customers a better user experience when searching for information on its sites or buying products related to its teams. Google invited its systems implementation partner CI&T to help execute the program.
Numerous challenges presented themselves across scoping, timing, and resourcing. What started out initially as a migration of around 1,200 virtual machines (VMs) turned into a much larger program comprising two new product launches in the cloud—with very fast turnaround requirements. CI&T originally scoped the program at a one-month cloud setup with a three-month migration. But that timing was quickly accelerated as the training season was just around the corner, and with it a large influx of traffic to their websites. Time was of the essence.
There were resourcing challenges as well. Initially, internal GCP technical expertise within the league’s staff was limited, as its infrastructure team was small and relatively new. Additionally, the existing tech teams already had multiple projects they were committed to across the organization.
Amid these challenges, CI&T stepped in and provided its Cloud Migration, DevOps, and GitOps expertise to automate various infrastructure-related processes as well as deploy new applications to the cloud. CI&T supported the different internal and external development teams involved in delivering applications and managing the operation of those applications in their multiple environments throughout the development lifecycle.
Helping them with infrastructure automation, monitoring, and operations perspectives was just the beginning. We also helped their development teams make the most of the Cloud by giving them continuous integration and continuous deployment tools that reduce the development cycle and deploy new features and fixes to production faster.
The e-commerce platform that was initially requested to be provisioned using Google Compute Engine Virtual Machines quickly evolved its microservices and was migrated to a more powerful and cost-effective alternative: Google Kubernetes Engine (GKE). Other Google Cloud Platform services such as Google Cloud SQL managed database service and Google Cloud Storage for the content were also used. For the CMS, given the more complex architecture, a multi-regional deployment of Google Container Engine (GKE) and multi-regional deployments of MongoDB were also created.
1. Improved customer experience for millions of fans
The migration and implementation was a win not only for the league, but for its customers as well. It translated into a seamless, always-on user experience, which was realized on the league’s opening day and weekend as millions of visitors flooded the sites. Over that weekend, the new system was able to seamlessly handle over 300 million requests within the CMS without a hitch, as well as 70 million e-commerce requests—with no data loss or business interruption. Subscription renewals for the league's TV networks, totaling $50 million in revenue, also went smoothly before the opening day, thanks in part to the new infrastructure.
CI&T leveraged all the many benefits that GCP affords in building an auto-scalable and highly available and self-correcting infrastructure, with the greatest technical impact being the ability to easily distribute complex workloads across the US for reliable, automatic failover capability.
2. Accelerated delivery
The professional sports league was able to deliver their first live project on GCP in less than eight weeks and we had the e-commerce segment live with all of the underlying infrastructure of the environment in the development lifecycle fully automated. Within the eight week period, the distributed development teams were able to publish their changes, tear down and rebuild environments, and control replication and autoscaling configurations in a very simple manner, allowing for fast-paced development and short feedback cycles.
This migration was a win for all.