Rate Limiter

Question 1

What happens in a token bucket when all tokens are consumed?

Accepted Answer

New requests are immediately rejected (no queue) — When the bucket is empty the limiter immediately hard-rejects each incoming request. A throttling variant would queue or delay requests instead — but the token bucket as shown here returns a rejection with no queuing.

Question 2

A rate limiter's purpose is to…

Accepted Answer

Limit the number of requests per time window — It caps request rates to protect services from overload, abuse, and to enforce fair usage/quotas.

Question 3

The token bucket algorithm works by…

Accepted Answer

Refill tokens at a fixed rate up to a bucket capacity — Tokens accumulate up to a capacity; requests need a token. It allows short BURSTS (up to the bucket size) while bounding the average rate.

Question 4

You fire a burst of 8 requests when the token bucket has 3 tokens remaining. How many requests are allowed?

Accepted Answer

3 are allowed, 5 are rejected — Each request consumes one token. Only the first 3 requests find a token; the remaining 5 are rejected immediately because the bucket is empty.

Question 5

In the sliding window mode, once the 5-request limit is reached, when does capacity free up?

Accepted Answer

As soon as any earlier request ages out of the 5-second window — The sliding window tracks individual request timestamps. As each recorded request ages past 5 seconds, it is removed and its capacity slot becomes available — so recovery is gradual, not a sudden burst at a fixed boundary.

Question 6

At the default refill rate of 2 tokens per second, how long does it take to recover from 0 tokens to 4 tokens?

Accepted Answer

2 seconds — At 2 tokens per second, accumulating 4 tokens takes 4 ÷ 2 = 2 seconds. You can confirm this by exhausting the bucket with Burst (8x) and watching the badge count climb back up.

Question 7

When a client exceeds the rate limit, the standard HTTP response is…

Accepted Answer

A 429 too many requests — Status 429 signals rate limiting, often with a Retry-After header telling the client when to try again.

Question 8

Token bucket is preferred when you want to…

Accepted Answer

Allow bursts up to bucket size while bounding the average rate — Its accumulated tokens permit bursts up to the bucket capacity, which suits bursty but bounded traffic patterns.

Question 9

In a distributed system, rate limit counters are typically stored in…

Accepted Answer

A shared store like Redis — Per-server local counters would let a client exceed the global limit by hitting different servers; a shared store enforces a consistent limit.

Question 10

Rate limiting by API KEY or user ID (rather than IP) is better when…

Accepted Answer

Many users share a single IP address through a corporate proxy — IP-based limits unfairly group users behind shared NATs/proxies; per-key/user limiting ties quotas to accounts accurately.

Which limiter is this?

Test your knowledge

What happens in a token bucket when all tokens are consumed?

The short version

Interview Questions

References & Further Reading

Related topics

Love this visualization?