Connection pooling to upstream in Nginx - Time & Space Complexity

Time Complexity: Connection pooling to upstream

O(n)

Understanding Time Complexity

We want to understand how the time to handle requests changes when using connection pooling in nginx.

Specifically, how does reusing connections affect the work nginx does as requests increase?

Scenario Under Consideration

Analyze the time complexity of the following nginx configuration snippet.

upstream backend {
    server backend1.example.com;
    server backend2.example.com;
    keepalive 32;
}

server {
    location / {
        proxy_pass http://backend;
    }
}
    

This config sets up connection pooling with 32 keepalive connections to upstream servers.

Identify Repeating Operations

Identify the loops, recursion, array traversals that repeat.

• Primary operation: Handling each incoming request by reusing or opening connections to upstream servers.
• How many times: Once per request, but connection reuse reduces repeated connection setup overhead.

How Execution Grows With Input

As the number of requests grows, nginx reuses existing connections up to the keepalive limit, reducing repeated setup work.

Input Size (n)	Approx. Operations
10	About 10 request handlings, few new connections
100	100 request handlings, many reuse existing connections
1000	1000 request handlings, most reuse connections, few new setups

Pattern observation: The work per request stays mostly steady because connections are reused, not reopened each time.

Final Time Complexity

Time Complexity: O(n)

This means the total work grows linearly with the number of requests, but connection reuse keeps each request efficient.

Common Mistake

[X] Wrong: "Connection pooling makes handling requests constant time regardless of request count."

[OK] Correct: Each request still needs processing, so total work grows with requests, but pooling reduces repeated connection setup time.

Interview Connect

Understanding how connection pooling affects request handling time shows you can reason about real server efficiency and resource use.

Self-Check

"What if the keepalive value was set to 1 instead of 32? How would the time complexity change?"