Big Data Applications in Gaming

Solution: Gaming

Company Description

A global leader in the mobile and social gaming industry. The company uses Big Data to increase the use and monetization of its games.

Business Goal

Succeeding in the gaming industry isn’t child’s play. The gaming developer needed to understand player behavior to optimize its games so that players stay in their “flow zone” and keep coming back for more. The game company also had to better monetize its games and improve virality factors to attract more players.

Architectural Overview

The gaming company realized from the very start that Hadoop was the best platform for its Big Data requirements to analyze petabytes of granular player interactions and contextual data, both structured and unstructured, every month. Yet, the business had trouble with its internal efforts to realize the full potential of this Big Data platform, running into scalability, usability, and integration issues.

Key Challenges

High Volume and Velocity

Analyzing millions of users, billions of user events, and hundreds of metrics across dozens of titles proved problematic. Jobs would take too long to run or just simply time out. The issue was made worse during crunch time when marketing, product management, and analysts needed immediate insights critical to moving forward with game production decisions. Compute resources couldn’t keep up with peak demand.

Mashup Headaches

Gaming analytics involves a cauldron of data sources to obtain a broad understanding of players and their activities that will help identify the features gamers like, how to expose additional players to the game, and how to get players to keep playing. For example, the gaming company wanted to merge social interaction data with game logs to analyze conversations within a game. Or, to acquire users likely to stay and play, it needed to access US census data such as income and paying status, and player demographics in its CRM system.

Time Consuming Cluster Management

The gaming company began its Hadoop initiative as a proof-of-concept with a single cluster. As it moved into production, the implementation naturally grew in size, leaving the gaming company wanting a more automated approach to configuring, deploying, and managing its Hadoop clusters.

Qubole Data Service (QDS) in Action

High-Performance Auto-Scaling

Using QDS’ fully elastic Hadoop engine, the gaming company no longer hits a wall when processing large jobs or during peak demand. Auto-scaling automatically adds or removes compute resources based on actual usage to meet the Big Data volume and velocity requirements for granular gaming analytics.

Big Data Integration Made Easy

QDS tackled the gaming company’s tough structured and unstructured data mashup challenges using its dozens of pre-built connectors as well as its ability to build custom connectors. No changes to the company’s on-premise infrastructure were required to integrate on-premise, cloud, and social media data sources.

Cluster Management on Auto-Pilot

The game company lets QDS’ automated cluster management take care of all the work to configure, deploy and manage its clusters. Plus, the company now has less downtime due to a reduction in configuration errors and features such as auto-healing to replace failed instances.

Similar Stories

At MediaMath (founded 2007), we’ve built the tools that enable and empower this new breed of marketing professional. Read Case Study

Insightera provides a cutting-edge, real-time inbound marketing platform that accelerates B2B customer acquisition. Read Case Study