Question #649
A company operates a SaaS platform on AWS using Amazon API Gateway REST APIs integrated with AWS Lambda across multiple regions. They offer different subscription tiers, with premium customers allowed up to 5,000 requests per second using unique API keys. Premium users report receiving 429 errors during peak times, and logs show the Lambda functions are not being triggered. What is the most likely cause?
The Lambda functions' reserved concurrency is exceeded.
The regional Lambda concurrency limit is reached.
The API Gateway account-level throttling limit is exceeded.
The API Gateway per-client rate limit is too low.
Explanation
The correct answer is C. API Gateway has an account-level throttling limit (default: 10,000 RPS) that applies globally across all regions. If the total requests from all premium users exceed this limit, API Gateway throttles requests, returning 429 errors without invoking Lambda. Premium users are allowed up to 5,000 RPS each; if multiple users hit their limits simultaneously, the account-level limit is exceeded.
Other options are incorrect because:
- A/B: Lambda concurrency limits (reserved/regional) would allow API Gateway to trigger Lambda, which would then return errors. However, logs show Lambda isn't triggered, indicating throttling occurred earlier.
- D: Per-client rate limits are set to 5,000 RPS as per the subscription, so misconfiguration here is unlikely unless explicitly stated. The issue arises from the global account limit, not per-client settings.
Key Takeaway: API Gateway's account-level throttling is a global limit. When scaling multi-region SaaS applications, ensure this limit is adjusted to accommodate total expected traffic.
Answer
The correct answer is: C