Customer Case Study
About Adform
Adform is a global full-stack AdTech platform, offering a demand-side platform (DSP), supply-side platform (SSP), and data management platform (DMP). With nine data centers worldwide, Adform processes billions of ad events daily, handling real-time bidding (RTB), audience segmentation, and campaign optimization for leading advertisers and publishers. To be successful, Adform has to operate with millisecond speeds, as its customers depend on low-latency data access, real-time processing, and system uptime.
Challenge
Scaling real-time bidding and analytics
Adform’s initial architecture relied on Microsoft SQL Server and Cassandra, using batch-based processing for user profiles and campaign analytics. However, as data volumes surged, they encountered severe scalability, performance, and availability challenges. Their global infrastructure needed to handle 11M+ operations per second and process between 900K–1.8M bid requests per second, all while maintaining millisecond-level response times.
Adform needed a high-performance, highly available solution to support real-time bidding and campaign reporting without the limitations of batch processing and high operational overhead.
Scaling challenges with Cassandra
Expanding Cassandra was costly and complex, with 250+ nodes across multiple global data centers
Batch processing was operationally inefficient
Slow insights hurt campaign performance, causing 4+ hour delays, which ultimately limited real-time optimizations and decision-making.
Availability risks
Scaling required manual intervention, and repeated node failures caused delays
Solution
Migrating to Aerospike for real-time processing
Adform migrated from Cassandra and Microsoft SQL Server to Aerospike, redesigning its architecture to support real-time event processing, instant lookups, and simplified scaling.
With Aerospike’s predictable performance and auto-scaling capabilities, Adform eliminated downtime risks and simplified infrastructure management, enabling seamless growth without constant tuning.
Aerospike replaces batch processing with real-time key-value lookups.
Document-like collection data types (CDTs) efficiently store and update campaign metrics.
Kafka integration enables high-speed streaming event processing.
Edge-to-system of record architecture reduces strain on central databases and ensures low-latency decisioning.
Results
High performance, lower costs, and real-time insights
Aerospike transformed Adform’s infrastructure, enabling real-time campaign reporting, lower operational costs, and industry-leading uptime. Adform gained a future-proof, real-time architecture that can handle massive data volumes with low latency and high availability by eliminating batch bottlenecks, simplifying scaling, and reducing infrastructure costs.
50% reduction in server footprint
From 32-node Cassandra to 3-node Aerospike clusters per data center
99.999% uptime
Ensuring continuous ad bidding and analytics
4x performance improvement
Processing 11M+ operations per second
Real-time campaign reporting
Reducing data delays from 4 hours to seconds
Edge storage with Aerospike + Kafka
Optimizing data processing across global data centers
Additional resources
For a deeper understanding and more insights, explore these additional resources.
See more