Course contentsShow
← Cloud and Infrastructure
Lesson 265 of 2,9895. Load Balancing and Traffic ManagementFree lesson

NLB Performance Characteristics

Learn about NLB throughput, latency, concurrent connections, and why NLBs handle millions of requests per second.

NLB Performance Characteristics

What you'll learn: How Network Load Balancers achieve extreme performance and why they're built for speed and scale.

Why NLBs Are Performance Powerhouses

Network Load Balancers operate at Layer 4 (the transport layer), which means they route traffic based on IP addresses and TCP/UDP ports without inspecting the actual content of your requests. This simpler processing model enables extraordinary performance.

Key Performance Metrics

Throughput: NLBs can handle millions of requests per second while maintaining ultra-low latency. Because they don't parse HTTP headers or application data, they process connections much faster than Application Load Balancers.

Latency: NLBs typically add only microseconds to 100 milliseconds of latency. Think of it like a traffic cop who just waves cars through based on license plate color (IP address) versus one who stops to read a detailed manifest in every trunk (application data).

Concurrent Connections: NLBs excel at maintaining millions of simultaneous long-lived connections. This makes them perfect for gaming servers, IoT devices, or high-frequency trading platforms where every millisecond counts.

Why Such High Performance?

NLBs achieve these numbers because:

  • They preserve the client's source IP address
  • They use connection pass-through instead of connection termination
  • They avoid complex application-layer processing
  • They leverage highly optimized network-level routing

This is why you'd choose an NLB for real-time applications, UDP-based protocols, or extreme throughput requirements—scenarios where Application Load Balancers would become bottlenecks.

Key Takeaway: Network Load Balancers deliver millions of requests per second with ultra-low latency by operating at Layer 4 and avoiding application-layer inspection, making them ideal for high-performance, connection-intensive workloads.