Scaling Blockchain Infrastructure to Billions of Requests: A Cloud-Native RPC Platform

Unpredictable traffic, massive storage needs, multi-chain complexity, and strict uptime make blockchain RPC nodes a scaling nightmare, especially on bare metal, where overprovisioning becomes the default (and expensive) safety net. We replaced that model with a multi-region Kubernetes-first, GitOps-driven platform, now serving millions of RPC requests across 150+ blockchain networks and reducing costs up to 10x for certain workloads. 
In this talk, we'll share how we architected and operate our RPC Platform - a comprehensive solution that combines: 
- Smart Storage Management: Optimized PVC provisioning with "shallow" vs "deep" nodes to reduce storage costs by 60% 
- Multi-Chain Support: Unified Helm chart supporting running any kind of blockhain node type like Ethereum, Optimism, Arbitrum, Polygon, zkSync, etc. 
- GitOps at Scale: ArgoCD-based deployments managing hundreds of node groups across multiple environments 
- Bare Metal Integration: Ansible automation for provisioning K3s clusters on bare metal infrastructure 
This talk shares actionable patterns for building and running multi-tenant Kubernetes platforms at scale, highlights the unique operational challenges of stateful blockchain workloads, and breaks down our technical architecture with real-world production metrics.