Verve Migrates Ad Tech Platform to Google Cloud to Boost Scale and Cut Costs
Verve, through its demand-side platform Dataseat, migrated its ad tech infrastructure to Google Cloud to unify distributed systems and reduce costly cross-cloud data transfers. The migration enabled faster scaling, cost reduction, and improved operational efficiency for real-time bidding up to 1.5 million auctions per second, while supporting a privacy-first model without user tracking. The Dataseat team adopted a parallel deployment strategy on Google Kubernetes Engine (GKE) to ensure smooth transition with minimal downtime, then optimized performance using Axion C4A instances and BigQuery analytics. The company is experimenting with Google Cloud's Gemini AI model to advance AI-driven experimentation and build a scalable, privacy-first ad tech platform for future innovation.
- Organization
- Verve
- Industry
- Tech & Comms
- Location
- Germany
Reported outcomes
90 minutes
timeTime & speed
Strategic outcomes
Primary read
Use case focus
Showing 3 of 3
- 1Cloud migration
- 2Real-time bidding optimization
- 3AI experimentation
- Verve faced challenges with managing complex, distributed infrastructure across multiple clouds, resulting in high data transfer costs, duplicate systems, and brittle legacy reporting pipelines.
- Scaling real-time bidding while preserving privacy-first ad targeting demanded precise, low-latency, and scalable infrastructure.
- Maintaining continuous campaign operations during the migration with near-zero downtime was critical.
- The team designed a parallel infrastructure deployment on Google Kubernetes Engine using Terraform, connecting old and new environments via VPN to gradually shift traffic to Google Cloud.
- They replaced legacy reporting pipelines with BigQuery for improved reliability and reduced overhead.
- They benchmarked and deployed Axion C4A instances to optimize compute efficiency and reduce server count for in-memory key-value operations.
- Ongoing exploration of Google Cloud's machine learning ecosystem, including MLOps and Gemini AI for experimentation, supports advanced AI-driven optimization.
- Successfully migrated with only 90 minutes of planned downtime within six months, consolidating infrastructure and cutting cross-cloud transfer costs.
- Achieved ability to process 1.5 million ad auctions per second with faster scaling and improved platform operational efficiency.
- Reduced infrastructure costs by deploying Axion instances and improved reliability through BigQuery analytics.
- Positioned as early adopter and close collaborator with Google on emerging AI technologies including Gemini AI to accelerate ad tech innovation while maintaining privacy-first approach.
Architecture
Parallel infrastructure deployment on Google Kubernetes Engine (GKE) using Terraform, VPN connection for traffic routing, Axion-backed C4A instances for compute optimization, migration off legacy pipelines to BigQuery, and integration of Gemini AI models for future experimentation.
Sources & evidence1
AI-generated summary. Verify important details with the linked sources before relying on this case.