The Boston Celtics sit atop the eastern conference of the NBA but their success may have as much to do with their speed off the court as on it due to a major upgrade to its basketball IT analytics capabilities.

The story, as documented in an IT case study released by the team, illustrates how a near-overwhelming amount of data became more manageable and accessible once the team switched from in-house analysis to a cloud-based operation.

Data collection by the Boston Celtics goes back decades and was initially performed by a group of mathematicians and coders in a basketball analytics group. The data analysis is critical for developing a game plan, decisions about trades and selecting draft picks.

However, the amount of data became more complex and demanding over time as new data sources became available. Optical tracking, for example, became a tool to monitor the movement of players, referees and the ball simultaneously. It soon became clear that the size of in-house data analysis operations was becoming difficult to manage and required additional personnel and resources.

Compounding the problem was that data analysis had become as slow as an aging point guard. Data processing took 24 hours to complete because the Celtics were using a statistical processing platform that required large and costly servers despite long periods of downtime. The team was paying for memory, bandwidth and compute cycles that weren’t being used.

The Celtics realized that what was needed was a cost-effective solution that could deliver game data to coaches in real-time. The Celtics cleared the IT bench and added Mission Cloud as its technology partner to guide the team in making the switch to an AWS cloud operation.

“Mission Cloud fits in perfectly,” says Jay Wessland, chief technology officer for the Boston Celtics. “There’s the breadth of knowledge across AWS services and they can deep dive when we need them to so our people can focus on their core priorities, which are basketball and what data is trying to tell us about basketball. Our main focus is not managing IT infrastructure and we don’t want it to be.”

Mission Cloud came to the game with a playbook that would enable the move to the cloud and improve efficiency while also keeping costs low. Phase One was a “lift and shift” move to AWS, similar to the standard pick and role play on the court. This removed bandwidth constraints and on-demand computing with servers that shut down when not in use. The Celtics also reduced their operational burden with MS SQL Server databases in a scalable, high performance RDS-managed service.

Phase Two was basically a trade, with Windows servers exchanged for Linux, thereby eliminating Microsoft licensing fees that were a significant cost to the organization.

Phase Three involved automating and modernizing the Celtic’s internal IT infrastructure. Mission Cloud was keen to find the right tools for the task even if it was not native to AWS. For example, Mission Cloud chucked Microsoft SQL and replaced it with a non-native, cloud-based data warehousing platform that offered better performance. Mission Cloud also overhauled container deployment, migrating from EC2 instances to ECS Fargate, coupled with Apache Airflow for orchestration. Other improvements included automation for Docker images as well as security features in line with NBA requirements.

Phase Four focused on specific optimization for the Celtics. For example, IT operations were split into separate domains for marketing, and another for all other activities.

The Boston Celtics move to the cloud resulted in significant costs savings but the number that made coaches and analysts happiest was a 71% reduction in data processing time after Phase One with additional 15% improvement after Phase Two. This allowed coaches to focus on their work and not the cloud.

“In reality, the fact that they don’t need to think about it is what makes it so great,” says Wessland. “Anytime a coach needs to log on and wants to look something up, it’s just there and it’s working. It’s invisible and it should be invisible.”