AWS - Architectures

Question 1

How much memory can you allocate to a Lambda function?

Answer 1

128MB to 10,240MB
AWS uses this to calculate how much CPU to allocate

Question 2

How much disk space do Lambda’s get and where do you access it?

Answer 2

Start out with 512MB of storage which you can scale up to 10,240MB. Accessible at /tmp

Question 3

What 5 runtime environments are supported by Lambdas?

Answer 3

Python, Ruby, Go, C#, Node.js

Question 4

How are permissions typically granted to Lambda functions?

Answer 4

Via IAM Execution Roles

Question 5

How could you use Lambdas for serverless file processing?

Answer 5

Via S3 + S3 Events. For example the Upload Completed S3 event could trigger a Lambda function to apply watermarks to an image.

Question 6

How could you use Lambda Functions in database triggers?

Answer 6

In DynamoDB, your DB triggers can invoke Lamdba functions assuming you enable DynamoDB Streams.

Question 7

How could you use Lambda Functions to implement serverless CRON?

Answer 7

EventBridge or CWEvents can trigger Lamda functions on a scheduled interval.

Question 8

What are 3 architectural methods for invoking Lambdas?

Answer 8

• Synchronous Invocation
• Asynchronous Invocation
• Event Source Mapping

Question 9

What 4 destinations can asynchronous Lambda functions send successes/failures to?

Answer 9

• SQS
• SNS
• Other Lambdas
• EventBridge

Question 10

How do asynchronous Lambdas handle failures? What design considerations should you use?

Answer 10

• Failures past the retry threshold get sent to a DLQ (dead-letter queue) - either an SQS queue or an SNS topic.
• Lambdas let you configure 0, 1, or 2 automatic retries
• Your functions have to be idempotent

Question 11

What properties do Lambda Function Versions have?

Answer 11

• Each version contains both the code to run and the configurations for that function.
• Published versions of a Lambda Function are IMMUTABLE and have their own ARN.
• $Latest points at the latest version of a Lambda.
• Aliases, such as DEV, STAGE, or PROD can point at a specific version of a Lambda, and can be changed

Question 12

When would you use the Lambda Provisioned Concurrency configuration?

Answer 12

When you need to keep X Execution Contexts warm and ready to use, to avoid latency associated with cold starts.