Reliance Games, the digital arm of the Reliance Entertainment Group, was on a hunt for a robust analytical platform to boost its revenue, enhance data-driven decision making, improve product development, gain a competitive edge, as well as make the platform flexible to manage costs. The company’s games—Real Steel, Pecific Rim, Hunger Games, and WRB—are amongst the world’s top 10 games in terms of gameplay and revenue. With more than 600 game developers working for over 2,000 devices across various operating systems, the company was looking out for a vendor who could understand their business dynamics and products and provide an end-to-end real-time big
data analytics solution.
The company’s basic need was to analyze the products at the product development level to have a better understanding of daily active users (DAUs), number of downloads, user behavior, game performance, dropouts, reasoning for drop-outs, etc. For these insights, the company required descriptive, predictive, and prescriptive analytics for current as well as future products. Though there are many big data vendors in the market, the company was facing challenges in finding a vendor who could provide it a technical backbone and business intelligence on these products. So, it decided to build a system from scratch.
The company thus embarked on a journey to build an end-to-end analytical setup, including infrastructure as well as platform for analytics. Sayed Peerzade, Vice President-Technology, Reliance Big Entertainment & Reliance Entertainment-Digital elaborates, “Since we have taken the cloud route as a strategy, it has already served the purpose of elimination of possible CAPEX. Moreover, Hadoop and the subsequent analytical platform was built and managed by us, eliminating the license and platform vendor costs.”
BUILDING THE ANALYTICS PLATFORM
Since the company required real-time analytics, it wanted to build a proper infrastructure architecture which could support processing and analyzing the data volume at a lighting speed. However, collecting events from across the devices of the world itself was a challenge, apart from the huge volume of data flowing inside and storing this data.
To address this, the company formed a private cloud with virtualization technique with huge server farms, required network architecture, and developed Apache/Java module to collect the data simultaneously from devices across the globe. These servers collect the data in flat file format and then push the data to storage using the cron
jobs set at an appropriate time.
“In an industry like entertainment and global businesses like gaming, it’s rare to find the complete lifecycle projects on big data and that too on an opex model,” asserts Peerzade. He further adds, “We have developed Hadoop distributed ecosystem with Hbase and Hive system using Intel architecture to process the data using set rules of regression. For appropriate decision making, data visualization techniques like Tableau were used to get data in a presentable format for decision making.”
“This project supports the daily data volume of 37 mn and total events reaching 100 bn per game mark. 40 GB data is being added every day and nearly 300 events are captured from each device daily for each game session,” shares Peerzade.
The success of this project brought tremendous advantages for the company. There has been a huge improvement in customer satisfaction and outreach, which means more revenue flowing in. The company has registered a significant increase in revenue after this project, owing to improved customer experience and quick insights on user behavior, enabling informed decision making. Apart from these, the overall quality of the product has also increased manifold, hence improving the user base, which is directly proportional to revenues.