Case Studies Global Gaming Giant Turns to Aerospike to Simplify its Architecture and Accelerate Growth
Edit This Case Study Record

Global Gaming Giant Turns to Aerospike to Simplify its Architecture and Accelerate Growth

Analytics & Modeling - Predictive Analytics
Application Infrastructure & Middleware - Database Management & Storage
E-Commerce
Software
Business Operation
Software Design & Engineering Services
System Integration
Playtika has six gaming studios, each with a different game. It was using a different NoSQL solution, which was not helping the company reach its goal of 150,000 reads and 50,000 writes. Further, the architectural limitations of that solution caused Playtika to implement multiple clusters of nodes per studio - something it did not want to persist. There were different workloads on different clusters (some were highly loaded and others were barely used) and it was difficult to scale out with such a configuration. Adding to the problems was that the incumbent NoSQL solution is not ideally built for SSDs because of its internal rebalancing process. That led to the disks being frequently exhausted, malfunctioning and forcing replacements every six to 12 months.
Read More
Playtika is a leading gaming company with 27 million monthly active users playing its titles. Founded in 2010, the company was among the first to offer free-to-play social games on social networks and, shortly after, on mobile platforms. Headquartered in Herzliya, Israel, the company has over two thousand employees in offices worldwide including Tel-Aviv, London, Berlin, Montreal, Chicago, Las Vegas, Santa Monica, Sydney, Buenos Aires, Tokyo, Vienna, Kiev, Bucharest, Minsk, Dnepr and Vinnitsa. On its website, Playtika explains that “We smash art together with science to bring the most engaging and fully customized games experiences. Our diverse portfolio of games is synchronized across platforms and devices.”
Read More
Playtika phased in one studio’s application first, refactoring it by changing the data model from document-based to key-value in order to better suit Aerospike. In addition, compression of stored data was added plus some operations will be moved to full-in-memory architecture. Three additional studios are planned for future migration. The overall results were significant: Cut server footprint reduced 6x – from 200 servers to 30 Aerospike servers. Boosted performance 300% - 200,000 reads and writes per second now 600,000. Grew data 70% - each studio was able to grow from 2TB of unique data to 3.4 TB. Simplified clustering - each studio reduced their cluster counts from six down to one. Nodes per cluster slashed – five-to-six nodes per cluster dropped to three-to-five nodes per cluster.
Read More
Playtika was able to phase in one studio’s application first, refactoring it by changing the data model from document-based to key-value in order to better suit Aerospike.
Compression of stored data was added, and some operations were moved to a full-in-memory architecture.
Three additional studios are planned for future migration.
Cut server footprint reduced 6x – from 200 servers to 30 Aerospike servers
Boosted performance 300% - 200,000 reads and writes per second now 600,000
Grew data 70% - each studio was able to grow from 2TB of unique data to 3.4 TB
Download PDF Version
test test