Warming up BigTable cluster

Jul. 1, 2019 // Linkedin // Twitter

This time we warmed up BigTable cluster (30 nodes in one location) by sending 523K updates per second using GoLang service. Our service consumed 270 vCPU and 10GB RAM (peak) to process 500K QPS (read from PubSub and send an update to BigTable).

What next? We asked Google to increase our BigTable limits (30+ nodes per location) - we want more requests.