Home › AWS Developer Associate › cloudcomputing › awsdevelopment › A Lambda function reads from a Kinesis Data Stre…
A Lambda function reads from a Kinesis Data Stream. Throughput is fine on average but spikes cause backlog. Which is the FIRST tuning step to reduce backlog WITHOUT changing the producer?
AMove the consumer from Lambda to a Snowball Edge appliance
BDisable the IAM execution role to reduce auth overhead
CDrop every other Kinesis record at the producer
DIncrease Kinesis Data Streams parallelisation factor (or batch size / number of shards) so more Lambda invocations can run in parallel
Answer & Solution
Correct answer: D. Increase Kinesis Data Streams parallelisation factor (or batch size / number of shards) so more Lambda invocations can run in parallel
Increase parallelisation (per-shard parallel batches), batch size, or shard count to scale the consumer side. Dropping records loses data; Snowball is offline transfer; disabling IAM isn't a real option.
Related questions
An API Gateway REST API in front of Lambda needs to authenticate requests using a Cognito Which strategy allows a Lambda function inside a VPC to access the internet (e.g., to callA serverless app needs to fan out a single event to multiple downstream Lambda functions, A Lambda function processes a large file from S3. Cold starts are unacceptably long on JavWhich AWS feature lets developers attach short-lived AWS credentials to a Lambda function What does it mean for an API endpoint to be IDEMPOTENT?A Lambda function is processing messages from an SQS queue. Some messages fail repeatedly.Which Lambda configuration setting most directly affects the function's CPU and network pe