-
Increase the configured CPU value
-
Increase the configured timeout value
-
Increase the configured memory value
-
Increase the configured concurrency value
-
blocks
-
layers
-
aliases
-
handlers
-
in sequence
-
both of these answers
-
neither of these answers
-
in parallel
-
aws lambda invoke --function ReturnBucketName outputfile.txt
-
aws lambda execute --function-name ReturnBucketName outputfile.txt
-
aws lambda invoke --function-name ReturnBucketName outputfile.txt
-
aws lambda execute --function ReturnBucketName outputfile.txt
-
AWS Trace
-
CloudStack
-
CloudTrail
-
AWS X-Ray
Q6. You need to build a continuous integration/deployment pipeline for a set of Lambdas. What should you do?
-
Create configuration files and deploy them using AWS CodePipeline.
-
Create CloudFormation templates and deploy them using AWS CodeBuild
-
Create configuration file and deploy using AWS CodeBuild
-
Create CloudFormation templates and deploy them using AWS CodePipeline.
-
API Gateway
-
S3
-
SAS
-
CLoudTrail
-
Use S3 metrics and CloudWatch alarms
-
Create custom metrics within your Lambda code.
-
Create custom metrics within your CloudWatch code.
-
Use Lambda metrics and CloudWatch alarms.
-
an SSL certificate
-
a bitmask
-
an AWS KMS key
-
an HTTP protocol
-
binaries.
-
all of these answers
-
executables
-
Shell scripts
-
MVC
-
virtual
-
stateless
-
protocol
-
by uploading a .zip file
-
all of these answers
-
by editing inline
-
from an S3 bucket
Q13. You are performance-testing your Lambda to verify that you set the memory size adequately. Where do you verify the execution overhead?
-
CLoudWatch logs
-
DynamoDB logs
-
S3 logs
-
Lambda logs.
-
CodeStack
-
ElasticStack
-
Mobile Hub
-
CodeDeploy
-
proportionally
-
equally
-
periodically
-
daily
Q16. You can restrict the scope of a user's permissions by specifying which two items in an IAM policy?
-
resources and users
-
resources and conditions
-
events and users
-
events and conditions
-
logging streams
-
rotating streams
-
logging events
-
advancing log groups
-
create a Lambda
-
be an event source
-
assign an IAM role
-
delete a Lambda
-
Create a Lambda function with a custom runtime and reference the function in your Lambda
-
Create a Lambda layer with a custom runtime and reference the layer in your lambda
-
You cannot use Lambda in this situation
-
Create a Lambda function with a custom runtime
-
the execution policy
-
the Lambda configuration
-
the Lambda nodes
-
the IAM user
-
department:Sales,department:Sales
-
department:Sales,department:sales
-
aws:demo;aws:demo
-
aws:demo;aws:DEMO
-
neither of these answers
-
UDP/IP
-
TCP/IP
-
both of these answers
-
automatically
-
none of these answers
-
manually
-
ad hoc
Q24. You are testing your stream-based application and the associated Lambda. AWS best practice advises you to test by varying what?
-
stream and record sizes
-
stream and shard sizes
-
batch and record sizes
-
batch and shard sizes
-
Place each subnet in a VPC. Associate all subnets to your Lambda.
-
Place all subnets in a VPC. Associate all subnets to your Lambda.
-
Configure your Lambda to be available to multiple VPCs.
-
Configure all application VPCs to be peered.
-
number of function calls
-
amount of code run
-
compute time
-
amount of infrastructure used
-
Author a Lambda from scratch.
-
Use a blueprint.
-
Use a .zip deployment package.
-
Use the serverless app repository.
-
/tmp
-
/default
-
/temp
-
/ds
-
Delete the function.
-
Set the function concurrent execution limit to 0 while you update the code.
-
Reset the function.
-
Set the function concurrent execution limit to 100 while you update the code.
-
Overprovision memory to run your functions faster and reduce your costs. Do not overprovision your function timeout settings.
-
Overprovision memory and your function timeout settings to run your functions faster and reduce your costs.
-
Do not overprovision memory. Overprovision your function timeout settings to run your functions faster and reduce costs.
-
Do not overprovision memory. Do not overprovision your function timeout settings to run your functions faster and reduce costs.
-
removing log groups
-
none of these answers
-
creating log groups
-
updating log groups
-
DynamoDB tables
-
key-value pairs
-
S3 buckets
-
none of these answers
Q33. You need to use a Lambda to provide backend logic to your website. Which service do you use to make your Lambda available to your website?
-
S3
-
API Gateway
-
X-Ray
-
DynamoDB
Q34. You are creating a Lambda to trigger on change to files in an S3 bucket. Where should you put the bucket name?
-
in the Lambda function code
-
in a Lambda environment variable
-
in the Lambda tags
-
in another S3 bucket
-
Deploy the Lambda
-
Export the function
-
none of these answers
-
Configure a test event
-
Fleece
-
NPM
-
none of these answers
-
Pod
-
CloudTrail
-
CloudWatch
-
CloudFormation
-
LogWatch
-
a table definition
-
queue isolation
-
STS Write
-
an SNS topic
Q39. You need to set an S3 event trigger on your Lambda to respond when data is added to your bucket from another S3 bucket. Which event type do you configure?
-
POST
-
"All object create events"
-
PUT
-
COPY
-
Lambda configuration from logging code
-
Lambda handler from logging code
-
Lambda handler from core logic
-
Lambda configuration from core logic
-
YAML definition
-
CloudFormation stack configuration
-
SAML deployment stack
-
Zip file of all related files
-
only at creation
-
only before deployment
-
never
-
anytime via configuration
- SAM templates are a superset of CloudFormation templates. SAM templates include additional resource types.
- SAM templates have some overlap with CloudFormation templates. Both SAM and CloudFormation templates include resource types that are not in the other type of template.
- CloudFormation templates are a superset of SAM templates. CloudFormation templates include additional resource types.
- SAM templates are a different name for CloudFormation templates. Both template types include the same resource types.