To estimate the maximum concurrant connections Nginx can handle, use the following formula:

(Total RAM in GB × 1024 − System Overhead) / Average Request Size

• Total RAM in GB: Physical memory available on the server.
• System Overhead: Memory reserved for OS processes, background services, and buffer space (typically 2–4 GB).
• Average Request Size: Size of each request in KB for static content (e.g., images, CSS) or MB for dynamic content (e.g., PHP, API responses).

On a 32 GB RAM server with 16 CPU cores, assuming 4 GB system overhead and an average static request size of 5 KB, the theoretical maximum concurrency is approximately:

(32 × 1024 − 4096) / 5 ≈ 5,898

In practice, such a system can sustain 40,000–60,000 concurrent connections before hitting file descriptor limits or network stack constraints.

Load Testing with Apache Benchmark (ab)

Install the benchmarking tool:

yum install httpd-tools -y

Run a basic test:

ab -n 2000 -c 200 http://127.0.0.1/index.html

• -n: Total number of requests to simulate.
• -c: Number of concurrent clients.
• -k: Enable HTTP keep-alive (optional).

Sample output:

Server Software:        nginx/1.12.2
Server Hostname:        127.0.0.1
Server Port:            80

Document Path:          /index.html
Document Length:        19 bytes

Concurrency Level:      200
Time taken for tests:   1.013 seconds
Complete requests:      2000
Failed requests:        0
Total transferred:      510000 bytes
HTML transferred:       38000 bytes
Requests per second:    9333.23 [#/sec] (mean)
Time per request:       101.315 [ms] (mean)
Time per request:       0.507 [ms] (mean, across all concurrent requests)
Transfer rate:          491.58 [Kbytes/sec] received

• Requests per second: Throughput metric — higher values indicate better performance.
• Time per request: Average time for a client to receive a resposne.
• Transfer rate: Network bandwidth utilization; helps detect bottlenecks.

Monitoring Active Connections

View TCP connection states for Nginx/Apache:

netstat -n | awk '/^tcp/ {++S[$NF]} END {for(a in S) print a, S[a]}'

Typical output:

ESTABLISHED 1597
TIME_WAIT 1057
SYN_RECV 30
FIN_WAIT1 51
FIN_WAIT2 504
LAST_ACK 5

Connection state definitions:

• ESTABLISHED: Active, ongoing data transfer.
• TIME_WAIT: Connection closed, waiting for cleanup (normal after high traffic).
• SYN_RECV: Incoming connection request received but not yet acknowledged.
• FIN_WAIT1/2: Connection closing in progress.
• LAST_ACK: Final acknowledgment pending before full closure.

Count active Nginx worker processes:

ps -ef | grep nginx | grep -v grep | wc -l

Count established connections on port 80:

netstat -antp | grep ':80' | grep ESTABLISHED | wc -l

Use these metrics to correlate system load with observed concurrency and adjust Nginx worker limits, keepalive timeouts, or file descriptor limits accordingly.