Burst and nodelay options in Nginx - Step-by-Step Execution

Process Flow - Burst and nodelay options

Request arrives

↓

Check token bucket

↓

Tokens available?

No→Queue request if burst > 0

↓

Queue full?

↓

Allow request

↓

Process request

↓

Next request

Requests check tokens; if none, burst allows queuing; nodelay controls immediate processing of queued requests.

Execution Sample

Nginx

limit_req_zone $binary_remote_addr zone=mylimit:10m rate=1r/s;
limit_req zone=mylimit burst=3 nodelay;

# Requests come in quickly

Limits requests to 1 per second, allows 3 extra bursts, and nodelay sends bursts immediately without delay.

Process Table

Step	Incoming Request #	Tokens Available	Burst Queue	nodelay Effect	Action Taken
1	1	1	0	nodelay on	Request allowed immediately, tokens now 0
2	2	0	0	nodelay on	Burst queue used (1), request allowed immediately
3	3	0	1	nodelay on	Burst queue used (2), request allowed immediately
4	4	0	2	nodelay on	Burst queue used (3), request allowed immediately
5	5	0	3	nodelay on	Burst full, request delayed or rejected
6	6	1 (refilled)	3	nodelay on	Token refilled, request allowed immediately
7	7	0	3	nodelay on	No tokens, burst full, request delayed or rejected

💡 Requests stop being allowed immediately when burst queue is full and no tokens are available.

Status Tracker

Variable	Start	After 1	After 2	After 3	After 4	After 5	After 6	After 7
Tokens Available	1	0	0	0	0	0	1	0
Burst Queue	0	0	1	2	3	3	3	3

Key Moments - 3 Insights

Why does the burst queue fill up even when tokens are zero?

Because burst allows extra requests beyond tokens to queue temporarily, as shown in steps 2-4 in the execution_table.

What does nodelay do when burst queue is used?

With nodelay on, queued burst requests are allowed immediately without delay, as seen in steps 2-4 where requests are allowed right away.

What happens when the burst queue is full and no tokens are available?

Requests are delayed or rejected, as shown in step 5 where burst queue is at max and tokens are zero.

Visual Quiz - 3 Questions

Look at the execution_table at step 3. How many requests are currently in the burst queue?

A 3

B 1

C 2

D 0

💡 Hint

Check the 'Burst Queue' column at step 3 in the execution_table.

At which step does the token bucket refill to allow a new request?

A Step 6

B Step 5

C Step 4

D Step 7

💡 Hint

Look at the 'Tokens Available' column for when it changes from 0 to 1.

If nodelay was off, what would happen to burst queued requests?

A They would be allowed immediately anyway

B They would be delayed and not allowed immediately

C They would be rejected immediately

D Tokens would refill faster

💡 Hint

Recall nodelay controls immediate processing of burst queued requests as shown in the execution_table.

Concept Snapshot

limit_req_zone defines request rate and token bucket.
limit_req with burst allows extra queued requests beyond tokens.
nodelay sends burst requests immediately without delay.
Without nodelay, burst requests wait for tokens.
Burst queue max size controls how many extra requests can queue.
Requests exceeding burst queue are delayed or rejected.

Full Transcript

This visual execution shows how nginx's limit_req with burst and nodelay options work. Each incoming request checks if tokens are available in the token bucket. If tokens exist, the request is allowed immediately and tokens decrease. If no tokens are available, burst allows extra requests to queue temporarily. When nodelay is on, these burst queued requests are allowed immediately without waiting. The execution table tracks tokens and burst queue size per request. When the burst queue is full and no tokens remain, requests are delayed or rejected. The variable tracker shows tokens and burst queue changes over time. Key moments clarify why burst queue fills and how nodelay affects request handling. The quiz tests understanding of burst queue size, token refill timing, and nodelay behavior. The snapshot summarizes the core nginx rate limiting concepts with burst and nodelay options.