Lambda

Question 1

What is Lambda?

Answer 1

A Function-as-a-Service capability for short running and focused processing.

Question 2

How are you billed for Lambda?

Answer 2

You are billed for the duration that a functions.

Question 3

What is a Lambda function?

Answer 3

A deployment package of function code and a runtime environment.

Question 4

What are the common runtimes for Lambda?

Answer 4

Python, Ruby, Java, Go, and C#. Custom runtimes are possible using layers.

Question 5

What is the range of memory size for Lambda?

Answer 5

128MB - 10240MB in 1MB steps.

Question 6

How much vCPU is allocated per MB of memory?

Answer 6

1vCPU per 1769MB of memory.

Question 7

What is the default amount of storage for Lambda and where is it located?

Answer 7

512MB of storage available as /tmp. This can scale up to 10240MB.

Question 8

What is the function timeout length for Lambda?

Answer 8

15 minutes (900 seconds).

Question 9

What type of networking are Lambda functions given by default?

Answer 9

Public networking. They can access public AWS services and the public Internet.

Question 10

What is a limitation of Lambda functions having public networking?

Answer 10

They cannot access VPC-based services unless the VPC-based services have public IPs and the appropriate security controls are in place to allow external access.

Question 11

What happens to Lambda functions that are running inside a VPC?

Answer 11

They obey all VPC networking rules. They cannot access public AWS services or the Internet without the appropriate services and security rules for public access.

Question 12

What are Lambda execution roles?

Answer 12

IAM roles attached to Lambda functions which control the permissions the Lambda function receives.

Question 13

What are Lambda resource policies?

Answer 13

A Lambda resource policy controls what services and accounts can invoke Lambda functions.

Question 14

What does Lambda use for logging?

Answer 14

CloudWatch Logs - Lambda execution output and logs
CloudWatch - Metrics such as invocation success/failure, retries, latency, etc.
X-Ray - distributed tracing of users and sessions

Question 15

What is synchronous invocation of a Lambda function?

Answer 15

The CLI/API or API Gateway invokes a Lambda function and waits for a response. The result (success or failure) is returned during the request. Errors and retries must be handled by the client.

Question 16

What is asynchronous invocation of a Lambda function?

Answer 16

Typically used when AWS services invoke Lambda functions. The service doesn’t wait for a response; the event is generated and the service stops tracking. On failure, Lambda will retry 0-2 times, but the function needs to be idempotent (reprocessing a results should have the same state).

Question 17

What is Event Source Mapping for a Lambda function?

Answer 17

Event Source Mapping polls streams or queues which don’t support event generation to invoke Lambda (Kinesis, DynamoDB streams, SQS).

Question 18

What happens if an asynchronous invocation of a Lambda functions fails too many times to assist in troubleshooting?

Answer 18

Events can be sent to a dead letter queue after repeated failed processing.

Question 19

What are Destinations with regards to asynchronous Lambda invocations?

Answer 19

Lambda supports sending successful or failed events to SQS, SNS, other Lambda functions, and EventBridge.

Question 20

What does the Event Source Mapping need to interact with the event source during execution of a Lambda function?

Answer 20

The Lambda execution role for the Lambda function.

Question 21

Do Lambda functions support versioning?

Answer 21

Yes. A version is the code and the configuration of a Lambda function. A version is immutable and each version has its own Amazon Resource Name.

Question 22

What variable points to the latest version of a Lambda function?

Answer 22

$latest

Question 23

What are aliases for Lambda functions?

Answer 23

Variables that point at different versions of a Lambda functions (i.e., DEV, STAGE, PROD).

Question 24

What is a cold start of a Lambda execution?

Answer 24

A cold start is a full creation and configuration of the execution context for a Lambda function, including function code download. It can take ~100ms.

Question 25

What is a warm start of a Lambda execution?

Answer 25

If a Lambda function is invoked again soon after its previous invocation, it may use the same execution context as the last invocation. The execution context creation can be skipped, reducing the startup time to ~1-2 ms.

Question 26

How to you plan to use warm starting of Lambda functions?

Answer 26

Provisioned concurrency will create and keep a specified number of execution contexts warm and ready for use. If you don’t use provisioned concurrency, you MUST assume you will get a new context each time. They might be reused, so you can try optimize, but you need to be prepared for a new context each time.