Skai Transformed Real-time Analytics at Scale with Twingo & SingleStore

About Skai

Skai (formerly Kenshoo) is a leading omnichannel advertising platform specializing in walled garden media. With nearly two decades of innovation, Skai empowers brands and agencies to run data-driven advertising programs across search engines, social networks, retail media, and app marketing. Backed by AI-powered decision-making and optimization solutions, Skai helps marketers drive impactful results with trusted data, and actionable insights.

Challenges

Today, Skai processes nearly a billion daily events. The rapid growth of publishers and their data volumes over the years presented several key challenges:

• Bottlenecks: The backend DB struggled with the increasing complexity of data. Queries involving tens of table-joins led to long response times of client requests.
• Real-Time Analytics: Batch processing was impacted too with client’s growth, the insights and optimizations were limited by the readiness of pre-processed data.
• Scale: Heavy clients needed special care, the old infrastructure wasn’t built for big data use cases.

These challenges highlighted the need for a transformative technological upgrade, for a scalable solution that could future-proof these pain points.

Solution

Skai partnered with Twingo to work on a long-term solution, the primary goal was to ensure real-time performance, scalability, and seamless user experience. The collaboration was tight and inclusive from the POC phase throughout full production deployment, integrating SingleStore as a caching layer on top of MySQL, enabling enhanced client experience with fast real-time analytics at any scale. Key Implementations:

• Tech Choice: The path to SingleStore, included a deep understanding of Skai’ needs, limitations and vision. Conducting a thorough POC with clear ROIs.
• Design: The solution required deep analysis on Skai’ workloads to derive the right Shard-keys, working methods and application adjustments. Twingo’s experts took an active part in the design and testing to ensure a good match.
• Optimizations: Custom hardware setups on VMs, private and public cloud helped Skai keep their performance optimal and cost effective.
• Real-Time Ingestion: The built-in Kafka pipelines enabled real-time replication from MySQL into SingleStore. By layering SingleStore as a caching layer and utilising its high DB performance, clients enjoyed fast UI response on their recent data.

Results

Skai transformed to a faster, more resilient and flexible platform, achieving dramatic improvements in performance and user experience. Key outcomes include:

• Responsive UI: Empowering client effectiveness with minimal idle-times.
• Scalability: Keep high-performance alongside growth of data and concurrency.
• Resiliency: Multi-tenant architecture and High-Availability features simplified the recovery process on Hardware outage.
• Advanced Capabilities: The adoption of unistore has set the stage for broader replacements of MySQL components. The advanced JSON features helped produce client flexibility for data enrichment of their choice.

Twingo’s expertise in real-time analytics and data architecture enabled Skai to provide unparalleled performance with their platform, ensuring they remain at the cutting edge of innovation in their industry.