Programmatic Advertising Optimization Platform

About the Client

The client is a programmatic advertising optimization company that operates a next-gen platform for publishers and advertisers. It aims to bring the supply and demand chains together and establish a direct connection between both parties.

The platform delivers value to advertisers and publishers by removing obstacles from media buying and selling processes, reducing mediators, and adding ultimate transparency and observability into day-to-day operations.

Goal

The client required qualified resources for their ongoing project in the shortest timeframe. The goal was to hire an experienced software engineer to optimize and maintain the oRTB’s underlying code, develop new features, and work on the system’s inner logic.

The client needed to meet critical performance requirements and take complete control over the speed of execution. Specifically, it was necessary to fit into the hard limits of 100,000 requests per 150-160 ml seconds and take active measures to prevent the system slowdown.

The client also needed to implement technology to identify and prevent traffic fraud from entering their programmatic platform.

Work Description

Geomotiv quickly provided a software engineer that became a part of the client’s in-house team.

As it was necessary to take full control of data processing speed and meet critical time limits, the developer dedicated much effort to code analysis, performance monitoring, and suggesting the best ways to optimize the oRTB’s system response time.

The platform ran on cumbersome Node.js code, which slowed down the execution speed. Our developer analyzed the code and suggested the most optimal optimization scenarios. For example, the developer took steps to simplify some parts of the underlying code and rewrote it using Golang. It helped reduce RAM usage and decrease the number of servers, among other things.

The project also required our specialist to monitor, identify, and fix memory leaks and other performance issues. To reach measurable results, our developer added metrics for analyzing ad delivery responses using Grafana and Prometheus.

The developer examined how the system behaved under heavy load as part of regular build activities. Our specialist ran load tests using the Locust framework. By pushing a small percentage of production traffic over the staging server and analyzing metrics, it was possible to measure the system’s speed and capacity and assess the response time.

As the client prioritized fraud detection, another critical task was implementing an anti-spam system that captured non-human traffic. This functionality enabled the client to prevent fraud from entering the oRTB flow and secure the ad-serving process.

Work Stages

1. Sprint planning
2. Implementation
3. Review and load testing
4. Release
5. Next Sprint planning

Results

Geomotiv’s software developer provided quality deliverables that became a solid ground for the client’s business success. Solutions suggested by the developer enabled the client to reduce costs and improve performance significantly.

Ongoing code optimizations and constant system monitoring helped to meet critical performance requirements. As a result, processing up to 100,000 requests per 150-160 ml seconds became possible, which was a primary concern for the client. Besides, the anti-fraud system also helped optimize and secure the ad-serving process and eliminate false traffic.