Weighted round-robin in Nginx - Step-by-Step Execution

Process Flow - Weighted round-robin

Start

↓

List servers with weights

↓

Select server based on weight

↓

Send request to selected server

↓

Next request -> Repeat selection

↓

Cycle continues

Requests are distributed to servers in proportion to their assigned weights, cycling through servers repeatedly.

Execution Sample

Nginx

upstream backend {
    server backend1.example.com weight=3;
    server backend2.example.com weight=1;
}

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

This config sends 3 requests to backend1 for every 1 request to backend2 in a repeating cycle.

Process Table

Request Number	Selected Server	Reason
1	backend1.example.com	Weight 3 means backend1 chosen first 3 times
2	backend1.example.com	Second request to backend1 due to weight
3	backend1.example.com	Third request to backend1 due to weight
4	backend2.example.com	Weight 1 means backend2 chosen once after backend1's 3
5	backend1.example.com	Cycle repeats, backend1 chosen again
6	backend1.example.com	Second request in new cycle to backend1
7	backend1.example.com	Third request in new cycle to backend1
8	backend2.example.com	backend2 chosen once again
9	backend1.example.com	Cycle repeats again

💡 The cycle repeats indefinitely, distributing requests according to weights.

Status Tracker

Variable	Start	After Request 1	After Request 2	After Request 3	After Request 4	After Request 5	After Request 6	After Request 7	After Request 8	After Request 9
Current Server	None	backend1.example.com	backend1.example.com	backend1.example.com	backend2.example.com	backend1.example.com	backend1.example.com	backend1.example.com	backend2.example.com	backend1.example.com
Request Count	0	1	2	3	4	5	6	7	8	9

Key Moments - 3 Insights

Why does backend1 get 3 requests before backend2 gets 1?

Because backend1 has weight=3 and backend2 has weight=1, nginx sends 3 requests to backend1 for every 1 to backend2, as shown in execution_table rows 1-4.

Does the order of servers in the config affect the weighted distribution?

No, nginx uses weights to decide distribution regardless of order. The cycle repeats based on weights, not position, as seen in the repeating pattern in execution_table.

What happens after all weighted requests are sent?

The cycle restarts from the first server again, continuing the weighted round-robin indefinitely, as shown by requests 5-9 in execution_table.

Visual Quiz - 3 Questions

Look at the execution_table, which server handles the 4th request?

A backend1.example.com

B backend3.example.com

C backend2.example.com

D No server assigned

💡 Hint

Check the 'Selected Server' column for Request Number 4 in execution_table.

At which request number does the weighted cycle repeat?

A Request 4

B Request 5

C Request 3

D Request 6

💡 Hint

Look at execution_table rows after Request 4 to see when backend1 is selected again.

If backend2 weight changes to 2, how does the selection change for first 5 requests?

A backend1 chosen 3 times, backend2 twice

B backend1 chosen twice, backend2 twice

C backend1 chosen 3 times, backend2 once

D backend1 chosen once, backend2 4 times

💡 Hint

Weights determine how many times each server is selected per cycle; increasing backend2 weight to 2 means 2 requests go to backend2 per cycle.

Concept Snapshot

Weighted round-robin in nginx:
- Assign weights to servers in upstream block.
- Requests distributed proportionally to weights.
- Cycle repeats indefinitely.
- Syntax: server name weight=number;
- Higher weight = more requests.
- Useful for load balancing uneven server capacities.

Full Transcript

Weighted round-robin in nginx means sending requests to servers based on their assigned weights. For example, if one server has weight 3 and another weight 1, the first server gets 3 requests for every 1 request the second gets. This cycle repeats forever. The configuration uses the upstream block with server lines specifying weights. The execution table shows which server handles each request in order, cycling through according to weights. Variables track the current server and request count. Key points include understanding why servers get requests in proportion to weights, that order in config does not affect distribution, and that the cycle repeats indefinitely. Quizzes test understanding of which server handles which request and how changing weights affects distribution.