Building amazing games for millions of users globally
Reduction in development time to accelerate new feature releases
Faster analysis of in-game activity to increase gamer engagement
“We selected Databricks Data Intelligence Platform to efficiently manage large-scale data processing in one place. It would have been impossible to develop ML models using data from siloed data lake and data warehouse environments, so with Databricks we have been able to simplify and unify our data architecture, and accelerate time to market for ML solutions.”
Devsisters launched Cookie Run to instant success, with a recorded 2.9 million daily users in Korea and over 10 million downloads within 12 weeks of its launch — and currently, the franchise has more than 200 million users. As data continued to grow along with an increase in players, Devsisters’ on-premises platform had performance issues and was complex to manage as well as costly to maintain. The company needed a new data analytics platform that data platform engineers and data scientists could use to analyze big data more quickly and seamlessly. To address its challenges, the company implemented the Databricks Data Intelligence Platform, which stores, understands and analyzes all data types for downstream analytics and machine learning. With this solution, Devsisters has simplified data analysis, reducing development time by 30% and enhancing query performance by 200%. With this newfound velocity in data processing and analysis, their business teams are able to rapidly extract new insights and pivot strategic and product decisions to better enhance the gaming experience.
High maintenance costs and performance issues limit data-driven decision-making
When global entertainment and gaming app developer Devsisters launched Cookie Run in 2013, the game was an instant success. The game — in which gingerbread cookies run to earn points, overcome obstacles, or fight and escape enemies — recorded 2.9 million daily active users in Korea and reached over 10 million downloads within 12 weeks of its launch. Cookie Run is ranked the No. 1 free game on the Apple App Store in Korea. The latest iteration, Cookie Run: Kingdom, won the 2021 Korea Game Awards Grand Prize in the game character category. It was also a nominee for the People's Choice Award category at the 2021 Pocket Gamer Mobile Games Awards: the first for a Korean gaming company. Today, the franchise has 200 million global users.
Devsisters began running the game on Amazon Web Services (AWS) and started using Apache SparkTM to analyze customer profile and in-game activity data to make better product decisions. With a data-driven culture in place, analytics has become critical to Devsisters' ability to achieve the company's mission of providing the best experience to players around the world with superior technology, service and content.
"We actively analyze our data to support decision-making for various teams, such as developing marketing strategy for global game launches, determining next updates on contents, and calculating the size of server infrastructure," said Hyemin Kang, Head of Business Intelligence at Devsisters.
To accelerate data-driven decision-making within the organization, Devsisters formed a business intelligence (BI) team. However, team members familiar with SQL Server and traditional data warehouse environments encountered challenges with the native infrastructure. There were high costs associated with maintaining the infrastructure, and BI employees frequently encountered issues requiring support from software engineers. As a result, data platform engineers focused too much of their time on troubleshooting rather than on enhancements.
Devsisters tried to solve this by building its own SQL-based data querying environment based on Spark Thrift Server, but challenges persisted. Update and Delete queries did not work, and performance was too slow, creating the possibility of lost data. Devsisters knew it would eventually be in a situation where it couldn't fully leverage its data to effectively support the business.
Databricks Data Intelligence Platform unlocks the promise of gamer data
To enable data warehouse engineers and data scientists to analyze big data easily and quickly, Devsisters needed a new data analytics platform that would unify all of its data and analytics. Its solution: the Databricks Data Intelligence Platform on AWS.
With the Databricks Data Intelligence Platform, Devsisters pivoted its focus from infrastructure maintenance to leveraging insights derived from players' data at its fingertips. "We manage a large scale of global businesses and work closely with various groups. A solution that integrates analytics, data engineering, BI, data science and machine learning is what we required, and Databricks offers just that," said SeungYong Oh, Head of Data Platform at Devsisters.
Devsisters also uses integrated capabilities like Databricks SQL, a serverless data warehouse on the Databricks Data Intelligence Platform that companies can use to run SQL and BI applications at scale with increased performance, while taking advantage of Delta Lake, an open source storage layer that increases data lake reliability and is compatible with Apache Spark APIs. "With the help of Databricks SQL and Delta Lake, we were able to engineer a data warehouse much more reliably. It is now possible to develop and manage a high-performance distributed big data warehouse at a low cost," explained Kang.
Additionally, Devsisters tested Photon, the next-generation engine on the Databricks Data Intelligence Platform that offers fast query performance at a lower cost, directly with actual data. "We conducted internal testing and were impressed that the performance of Photon was two times higher than expected," said Oh.
Driving gaming success with data-driven insights
By using Databricks Data Intelligence Platform to support the data warehouse, Devsisters' data warehouse engineers redesigned and loaded logical data in the data lake for decision-making. Business data analysts provide dashboards and analytics reports required for decision-making based on this data.
A game like Cookie Run: Kingdom processes more than 10TB every day, and data engineers can now process and analyze game player data 200% faster than before. With simplified data analysis and data warehouse management, data platform engineers are freed up to focus on improving data platform capabilities and performance. As a result, the data team is innovating faster, with development time reduced by 30%.
"Our query performance improvement means that decision-makers can access data more easily and gain business insights faster. Queries that used to take one minute now take under 20 seconds using the Databricks Data Intelligence Platform," said Kang.
Access to these better data insights has already started to pay off. Devsisters initially used the data lake to create a statistical model that predicts the lifetime value derived from the entire game process, from users entering the game to payment. However, the company decided it would be more appropriate to use data warehouses to maintain consistency with other indicators that are used for decision-making. The lakehouse provides an integrated data environment, combining the strengths of both a data lake and a data warehouse to enhance integration among indicators. The solution combines log ETL jobs, predictive modeling, data visualization and dashboards for business intelligence analysts, in one single platform. As a result, Devsisters can deliver powerful insights while providing reliable data.
Oh added, "It would have been impossible to develop models or efficiently manage large-scale data processing with a separate data lake and data warehouse solution. Databricks makes it easy to process all our data in one unified platform, and we can take advantage of this to focus on creating the best gaming experience for our customers."
