Performance Benchmarks

Built for Extreme Performance

Real-world benchmarks demonstrating industry-leading throughput and latency

Test Environment

All benchmarks performed on consistent hardware for accurate comparison

CPU

Intel Xeon E5-2686 v4

16 cores @ 2.3GHz

32 threads

Memory

32GB DDR4

2400 MHz

ECC Enabled

Network

10 Gbps NIC

Intel X540

Jumbo Frames

Ubuntu 22.04 LTS

Kernel 5.15

Tuned for performance

Performance Metrics

Measured with wrk2 benchmark tool, 1KB response payload

Throughput

1.2M

requests per second

Peak performance

Latency P50

87μs

median latency

Sub-100μs

Latency P99

890μs

99th percentile

Sub-millisecond

Latency P99.9

2.1ms

99.9th percentile

Concurrent Connections

1M+

simultaneous connections

Memory Usage

92MB

for 100K connections

Latency Distribution

Response time percentiles under 1M RPS load

P50 (Median)87μs

P75320μs

P90580μs

P99890μs

P99.92.1ms

Algorithm Performance

Throughput comparison across different load balancing algorithms

Requests Per Second by Algorithm

Higher is better - measured at 80% CPU utilization

Round Robin1,200,000 RPS

Weighted Round Robin1,180,000 RPS

Random1,195,000 RPS

Least Connections1,050,000 RPS

IP Hash1,120,000 RPS

URI Hash1,110,000 RPS

Scalability

Linear scaling with additional CPU cores and backend servers

Throughput vs CPU Cores

Near-linear scaling up to 16 cores

1 cores85K RPS

2 cores165K RPS

4 cores320K RPS

8 cores620K RPS

16 cores1.2M RPS

Memory Efficiency

Memory usage per 100K connections

100K connections92MB

250K connections215MB

500K connections420MB

750K connections625MB

1M connections830MB