How Bitdrift Managed 121 Million Concurrent gRPC Connections with AWS CloudFront and Route 53

Bitdrift, a mobile observability platform, scaled to handle over 121 million concurrent gRPC connections during the T20 World Cup by addressing DNS routing bottlenecks in their AWS CloudFront and Network Load Balancer infrastructure. Changing Route 53 routing policies from weighted to multi-value answers allowed smooth load distribution and zero server-side errors under extreme real-time traffic surges.

Weighted DNS routing concentrated connections on single NLBs under massive gRPC load.
Multi-value answer routing from Route 53 balanced traffic across multiple origins at DNS resolution.
Network Load Balancers paired with Amazon EKS workloads enabled scalable long-lived connection handling.

Infrastructure signal

Managing 121 million concurrent gRPC connections required rethinking DNS routing in front of AWS Network Load Balancers. Weighted routing returned a single IP address per DNS query, causing CloudFront edge nodes to direct all long-lived connections to the same load balancer for the duration of the DNS TTL. This resulted in a concentration of load and server errors despite increasing the number of NLB instances.

Shifting the Route 53 routing policy to multi-value answer routing changed the dynamic completely. It allowed each DNS query to return multiple IP addresses simultaneously, enabling CloudFront edge nodes to spread connections evenly among several NLBs from the outset. This upgrade to the routing layer was fundamental in achieving stable scaling and eliminating overload during sudden massive connection surges.

Developer impact

Developers supporting persistent connection workloads via gRPC must consider DNS routing strategies as a critical part of their deployment architecture. Traditional weighted routing strategies suited for stateless HTTP requests can lead to severe performance bottlenecks in use cases with long-lived connections, as DNS cache effects amplify uneven load distribution.

Bitdrift’s experience underlines the importance of aligning DNS configuration with connection patterns. Incorporating multi-value answer routing enables more predictable traffic distribution and reduces risk of cascading failures during peak traffic. Teams should integrate DNS routing checks and load balancer scaling considerations into their CI/CD pipelines and monitoring workflows to preempt these issues.

What teams should watch

Teams deploying real-time telemetry or other persistent connection workloads on AWS should evaluate their use of Amazon Route 53 routing policies. Weighted routing might cause unexpected congestion on single origins when CloudFront or similar CDNs are involved with long-lived connection protocols like gRPC.

Monitoring DNS resolution patterns, load balancer health, and connection distribution at the edge is essential. Observability should include metrics that highlight per-origin connection concentration and response errors. Evaluating multi-value answer routing compatibility with existing infrastructure and verifying health check configuration per DNS record are recommended to ensure resilience.

Additionally, developers must coordinate deployment scaling with DNS TTL configuration and load balancer IP count, as adding more origins without suitable DNS settings may not alleviate overload. This case demonstrates how small configuration changes at the DNS layer can dramatically improve cloud reliability and operational efficiency for demanding real-time services.

Source assisted: This briefing began from a discovered source item from AWS Architecture Blog. Open the original source.

How SignalDesk reports: feeds and outside sources are used for discovery. Public briefings are edited to add context, buyer relevance and attribution before they are published. Read the standards