What is Lambda Throttling? (and how to fix it!) | AWS Feature Overview

TL;DR

Learn how to prevent Lambda throttling in AWS applications.

Transcript

Read and summarize the transcript of this video on Glasp Reader (beta).

Key Insights

• Lambda throttling occurs when the concurrent execution count exceeds the concurrency limit, resulting in rejected requests.
• AWS Lambda has a default concurrency limit of 1,000 per region, which can impact multiple functions within the same account.
• Throttling can affect unrelated functions within the same account, causing cascading failures in different parts of an application.
• Different invocation event sources handle throttling differently, with SNS, SQS, and Kinesis having unique behaviors.
• Best practices to prevent throttling include requesting concurrency limit increases and setting reserved concurrency per function.
• Configuring a Dead Letter Queue (DLQ) is crucial for capturing failed invocations due to throttling and retrying them later.
• Alarming on throttling occurrences and DLQ messages is essential for timely intervention and mitigation.
• Implementing exponential back-off strategies can help manage throttling exceptions effectively.

Questions & Answers

Q: What is Lambda throttling and why does it occur?

Lambda throttling refers to the rejection of requests when the concurrent execution count exceeds the configured concurrency limit. It occurs to maintain system stability and prevent overloading. AWS Lambda has a default limit of 1,000 concurrent executions per region, affecting all functions in the account.

Q: How does throttling affect multiple Lambda functions in the same account?

Throttling affects multiple Lambda functions in the same account because the concurrency limit is shared across all functions. A sudden burst of traffic to one function can cause throttling for other unrelated functions, leading to cascading failures in different parts of an application.

Q: How do different invocation event sources handle throttling?

Different invocation event sources handle throttling uniquely. For example, SNS and SQS retries requests upon throttling, while Kinesis results in immediate failures. Configuring appropriate retry policies and using DLQs is essential to manage these behaviors and ensure message delivery.

Q: What are some best practices to prevent Lambda throttling?

Best practices to prevent Lambda throttling include requesting an increase in the concurrency limit, setting reserved concurrency for specific functions, and configuring DLQs. These steps help manage concurrency and ensure that critical functions have dedicated resources, reducing the risk of throttling.

Q: How can exponential back-off strategies help manage throttling exceptions?

Exponential back-off strategies help manage throttling exceptions by gradually increasing the wait time between retries. This approach reduces the load on the system, allowing it to recover before processing additional requests, thus minimizing the impact of throttling on application performance.

Q: Why is configuring a Dead Letter Queue (DLQ) important in managing throttling?

Configuring a DLQ is important in managing throttling because it captures failed invocations that exceed concurrency limits. DLQs allow these messages to be retried later, ensuring that critical data is not lost and providing a mechanism to handle failures gracefully.

Q: What role does alarming play in managing Lambda throttling?

Alarming plays a crucial role in managing Lambda throttling by notifying administrators of throttling events and DLQ message presence. This enables timely interventions to address issues, adjust configurations, or implement mitigation strategies, ensuring application stability and performance.

Q: How can one request an increase in the concurrency limit for AWS Lambda?

To request an increase in the concurrency limit for AWS Lambda, navigate to the AWS Support Center in the console and submit a technical request. Specify the desired concurrency limit increase, and AWS will review and approve the request, allowing more concurrent executions for your functions.

Summary & Key Takeaways

• Lambda throttling happens when concurrent executions exceed the set limit, causing request rejections. This video explains throttling, its causes, and how to handle it effectively.

• The video discusses the default concurrency limit of 1,000 per region and how it can impact multiple Lambda functions within the same AWS account.

• Best practices to mitigate throttling include increasing concurrency limits, setting reserved concurrency, and using DLQs to manage failed requests.