The Tech Behind the Bet: Low-Latency, High‑Scale Sportsbook on GCP

In sports betting, every millisecond matters. Odds must update instantly, wagers must confirm without delay, and the platform must handle massive bursts of traffic - all while staying compliant with regulations across multiple jurisdictions. 

A goal in the Champions League final, a knockout in a UFC fight, or a decisive ace in tennis can trigger thousands of bets within seconds.
Delasport’s solution: a Google Cloud-powered DevOps architecture built for low latency, high throughput, and rock-solid reliability.
The company’s Senior Infrastructure DevOps Architect Georgi Lazarov walks us through the process of achieving this remarkable result. 

Architectural Pillars 

Global Low-Latency Delivery
Delasport’s strive for close to zero latency makes the team explore every tech advancement out there and utilize those that would improve the processes significantly. This is why the list of services, solutions, and approaches is rather impressive itself: 

• Cloudflare Edge: Argo Smart Routing shortens request paths and avoids congestion, delivering up to 30% latency improvement for users far from GCP regions. 
• Cloudflare Caching: Serves static and semi-dynamic sportsbook content directly from the edge, reducing backend hits by 40-60% during high-traffic periods. 
• Modern Transport: Brotli (br) compression over TLS 1.3 reduces payload size by 15-20% compared to gzip and speeds up secure handshakes by up to 40%. 
• Global Reach: GCP Global External HTTP(S) Load Balancer (Anycast IP) with container‑native Network Endpoint Groups (NEGs). Anycast IP brings users onto Google’s backbone close to the edge, zonal NEGs route directly to pods in the nearest zone/cluster, cutting hop latency by ~10–15% 
• Placement: Topology-aware routing (PreferClose) keeps users on the nearest healthy zone, reducing cross-zone latency by 20-35%. 
• DNS Optimizations: CoreDNS caching reduces DNS lookup times by 50-70% inside the cluster. 

All of this and more helps Delasport’s solutions instantaneously reach every part of each regulated market they’re present in.

Elastic Scalability (Without Sacrificing Speed)

Just as important is the scaling stack that Delasport’s infrastructure team uses to ensure operators’ flexibility. It includes some of the best solutions in IT in general:

Cluster Scale: GKE Cluster Autoscaler with Node Auto-Provisioning (NAP) scales node capacity within minutes, enabling up to 5× throughput growth without latency degradation.

Ingress Scale: KEDA dynamically scales the ingress controller based on RPS/latency, reducing 99th percentile response time by 20% during traffic bursts.

Application Tier Scale: KEDA scales frontend services on live traffic metrics, keeping p95 latency under 150ms even at peak load.

Guardrails: PodDisruptionBudgets, topology spread, and slight over-provisioning prevent brownouts during scale events.

Scaling for Peak Moments

During the World Cup final, Delasport sustained tens of thousands of transactions per second with low p95 latency because:

• Cloudflare Argo combined with caching, Brotli, and TLS 1.3 cut end-to-end latency by up to 30% and reduced backend hits by 50%.
• Container-native load balancing (NEGs) routed traffic directly to pods in the nearest zone, lowering hop latency by 15%.
• KEDA scaled frontend and ingress in real time, keeping p99 under 300ms at 5× normal traffic.
• Cluster Autoscaler + NAP expanded compute capacity just-in-time across multiple zones.
• DNS queries were served from in-cluster cache, improving lookup performance by 70%.

Lessons Learned

From trying out the latest innovations come some key lessons. The DevOps Engineers and Infrastructure Experts at Delasport know this all too well. This is why they make sure to keep track of the important takeaways.

The first lesson in their textbook says that low latency isn’t just about speed - it’s the foundation of scalability and reliability in real-time sports betting.

Another important point is that by combining Cloudflare’s edge optimizations (Edge Network, Argo Smart Routing) with Google Cloud’s global networking (Global External HTTP(S) LB with Anycast IP plus zonal NEGs ), Topology-Aware (PreferClose) routing, DNS tuning, GKE Cluster Autoscaler (with NAP), and KEDA‑driven autoscaling, they’ve built an infrastructure that adapts instantly to spikes. At the same time it keeps the player experience smooth and secure.

“When the next big sporting moment happens, we’ll be ready - instantly, securely, and globally”, Georgi Lazarov says in conclusion.

Key Takeaways

• Latency is Everything: Low latency is not just about speed - it’s the backbone of a scalable, reliable sportsbook that can handle real-time betting events without hiccups.
• Traffic Optimization Path: The combination of Cloudflare’s edge network, Argo Smart Routing and Google Cloud’s container-native load balancing delivers a 30-40% latency reduction globally.
• Autoscaling Done Right: GKE Cluster Autoscaler, Node Auto-Provisioning, and KEDA allow Delasport to scale up to 5× normal throughput in minutes - without sacrificing response times.
• Peak-Proof Performance: During events like the World Cup final, Delasport kept p95 latency under 150ms and p99 under 300ms, even at massive traffic spikes.
• Continuous Innovation: The team actively experiments with new networking and scaling technologies, turning lessons learned into production-grade solutions.
• Global-First Mindset: Topology-aware routing, DNS caching, and optimized transport protocols ensure users worldwide enjoy the same fast, smooth betting experience.

All of this makes Delasport’s Infrastructure vertical one of the most intriguing in the industry and what’s even better - they’re always looking to grow, so Georgi advises everyone to keep an eye on Delasport’s career pages.