A startup swapped Supavisor and PgBouncer for PgDog on EKS. The swap stopped serverless deploy connection spikes. A multi-threaded, colocated pooler handled the bursty traffic.
PgDog needed fixes for Prisma prepared-statement handling. The team shipped those. PgDog now exports metrics via OpenMetrics to Prometheus/Grafana. It supports health-aware load balancing. That combo let the startup shrink Supabase hosts and cut costs.
