Sportradar's radar360 Powers Real-Time Sports Analytics Using Citus
Delivering a lightning fast data dashboard to teams and media
If you’re a hardcore sports fan and watch live games on TV or listen in on radio, you know how important stats are in telling the story. For professional commentators it’s absolutely critical that the stats and analysis is fed to them in the moment in milliseconds and not seconds later. Imagine watching a game and only finding out after the fact that a historic record was broken or that a critical milestone was reached - it’s too late!
Now think about how quick and responsive the data analytics engine needs to be to enable that commentary. To get these key data points, leading media companies and professional sports teams rely on Sportradar. When an analyst needs to pull stats on the fly, it uses the Sportradar radar360 dashboard. From every stat and score, to scenario based analysis like who has the best passer rating when they are down by 3 with 2 minutes left in the game in their opponent's red zone; can be pulled on the fly to match the live action and boost the entertainment value for the viewer.
To power the analytics dashboard, radar360 needed a highly performant, scalable and flexible database. They found this in Citus and couldn’t be happier. To learn more about how Sportradar solved its data scale and responsiveness problems, we chatted to Sportradar Head of Product Development, Michael Moser.
For Sportradar there are two kinds of customers. One kind are professional teams, that use the radar360 dashboard to track player performance, evaluate up and coming matchups and promote their team through their PR departments. They also use the powerful data analytics to strategize before games. The other kind are media outlets that broadcast live games. Especially for media, getting instant analytics matters.
Sportradar is not only a leader in data. They recognize how a product like radar360 can enhance story-telling by bringing insightful analysis to the fans.
“We’re in the storytelling business, not just the data business.”Michael Moser, Head of Product Development, Sportradar
Sports data - stat
When Sportradar was investigating how to build the new radar360 product, the most important criteria were ad-hoc dynamic queries, database speed and scalability. The nature of sports and fan experience means that getting data and analysis needed to be in milliseconds.
When considering their database solution, Sportradar evaluated Amazon Redshift data warehouse but "Redshift wasn't the right fit for a real-time web based dashboard with ad-hoc queries that need to return in sub-second times. The upfront execution plan compile time for each new ad-hoc query was too slow/variable. Citus' real-time execution did not have that overhead and allowed us to have scale, speed and dynamic ad-hoc queries." Said Moser.
Citus proved to be the best choice for Sportradar to keep using PostgreSQL and also scale their engine with more data while allowing real-time analytics. This is why Citus’ sharding and parallelization ability was so attractive to Sportradar. During testing, Citus proved to be up to 36 times faster using distributed nodes, compared to one node running on PostgreSQL.
In the future, Sportradar plans to increase the amount of data stored by 2-3X, with more historical data sets. Moreover, the future of sports strategy will rely on more statistics. Knowing this, Sportradar plans to add next-gen stats which generate information on chips placed on players. This will increase their data volumes even further, up to 10 times. Since Citus can scale horizontally as data grows by adding more hardware, Sportradar is confident about these exciting opportunities.
Winning the game: faster time to market
Beyond being insanely fast, another huge benefit for Sportradar from using Citus has been the time to get to market. radar360 is a brand-new product for the company, so it was important to choose a database solution that was quick to implement and not require large scale application changes. The engineering team was also happy knowing they would always be on the most up to date version of PostgreSQL, since the Citus extension isn’t a fork.
With radar360, Sportradar is well positioned to provide world class analytics to the more than 40 sports it covers across the globe.
So the next time you’re watching a nail-biting game and see the stats pop up, remember Sportradar and Citus is making it happen.