Types: Nginx Load Balancing

Nginx supports multiple load-balancing algorithms to distribute traffic among backend servers. By default, it uses Round Robin, but you can change it based on your needs.

---

Default: Round Robin (No special config needed)

How it works?
Requests are sent to each server in order, one by one.

Example:

• Request 1 → server 1 (3000)

• Request 2 → server 2 (3001)

• Request 3 → server 3 (3002)

• Request 4 → server 1 (3000)

• … (Repeats)

Example Nginx Config (Default Round Robin)

upstream my_api {
    server 127.0.0.1:3000;
    server 127.0.0.1:3001;
    server 127.0.0.1:3002;
}

server {
    listen 80;
    location / {
        proxy_pass http://my_api;
    }
}

Simple and fair distribution.
No extra configuration needed.
Doesn’t consider server load or response time.

---

Least Connections (Good for uneven load)

How it works?
Requests go to the server with the least active connections. This is useful if some requests take longer than others.

Example:

• If server 1 is handling 5 requests but server 2 has 2, Nginx will send the next request to server 2.

Nginx Config

upstream my_api {
    least_conn;
    server 127.0.0.1:3000;
    server 127.0.0.1:3001;
    server 127.0.0.1:3002;
}

Helps when some requests take longer.
Good for APIs with variable response times.
May cause imbalances if one server is much faster.

---

IP Hash (Stick same user to same server)

How it works?
Requests from the same IP always go to the same backend server. This is useful for session-based apps (e.g., authentication, shopping carts).

Example:

• User A (IP: 192.168.1.1) → server 1 (3000) always.

• User B (IP: 192.168.1.2) → server 2 (3001) always.

Nginx Config

upstream my_api {
    ip_hash;
    server 127.0.0.1:3000;
    server 127.0.0.1:3001;
    server 127.0.0.1:3002;
}

Keeps user sessions consistent.
Useful for apps with login sessions.
If a server goes down, users lose access (no automatic failover).

---

Weighted Load Balancing (Prioritize faster servers)

How it works?
Assigns weights to servers. A server with a higher weight gets more requests.

Example:

• server 1 (3000) → weight 3 (gets 3x more traffic).

• server 2 (3001) → weight 2.

• server 3 (3002) → weight 1 (gets the least traffic).

Nginx Config

upstream my_api {
    server 127.0.0.1:3000 weight=3;
    server 127.0.0.1:3001 weight=2;
    server 127.0.0.1:3002 weight=1;
}

Distributes load based on server capacity.
Great if some servers are more powerful.
Needs manual tuning.

---

Which One Should You Use?