AWS Messaging (SQS, SNS, Kineses)

Question 1

What is the default and max retention of messages in SQS?

Answer 1

4 days is the default, 14 days is the maximum

Question 2

How big can SQS messages be?

Answer 2

Up to 256KB

Question 3

True or False: SQS messages are deleted once consumed

Answer 3

False. The polling application must use DeleteMessage API to delete messages

Question 4

Why would you use SQS?

Answer 4

To decouple applications from one another, to turn synchronous process async

Question 5

We want to auto-scale our EC2 instances to handle more SQS messaging. What can we do?

Answer 5

We can create a CloudWatch metric on Queue length. We can then create a CloudWatch alarm that our ASG picks up to scale up our Instances

Question 6

We want to limit who and what resources can write to SQS. What can we use?

Answer 6

SQS Access Policies

Question 7

What is the SQS Message Visibility Timeout?

Answer 7

When a message is polled by a consumer, the message becomes invisible for a duration of time (e.g. it cannot be polled by other consumers)

Question 8

An application is processing an SQS message but knows it cannot finish within the Visibility Timeout window, what can it do?

Answer 8

Hit the ChangeMessageVisibility API to get more time

Question 9

What is SQS Long Polling?

Answer 9

When a consumer requests a message from the queue, if there are none in there, it can wait longer for messages to arrive through Long Polling

Question 10

How can we make sure SQS messages are in order?

Answer 10

Enable a FIFO Queue

Question 11

What is the SNS + SQS Fan Out Pattern?

Answer 11

Its where you use an SNS Topic to publish to multiple SQS’s so that individual applications/services have copies of their own messages. SQS also make sure we have no data loss with as retries, delays and data persistence

Question 12

What must we do to make sure SNS can publish to SQS?

Answer 12

Modify the SQS access policy

Question 13

We want to control what messages can be sent from our SNS Topics to our subcribers. What can we use?

Answer 13

We can use a JSON message filtering policy

Question 14

What does the Kinesis Data Stream record consist of?

Answer 14

A sequence number, a partition key and data blob (up to 1 MB)

Question 15

What is the difference between on-demand and provisioned capacity mode for Kinesis Streams?

Answer 15

On-demand mode allows you to scale automatically the number of shards in the data stream. Provisioned is specifying beforehand the amount of shards you would like

Question 16

What is the use case for Kinesis Data Firehose?

Answer 16

Firehouse will write data for us automatically to destination services

Question 17

Which one supports replay capability: Kinesis Firehose or Streams?

Answer 17

Streams

Question 18

What is the purpose of the partition key in a Kinesis Stream record?

Answer 18

It routes the record to different shards of a data stream

Question 19

How do you scale SQS?

Answer 19

SQS scales automatically

Question 20

Does SQS visibility timeout prevent other consumers from processing the message?

Answer 20

Yes

Question 21

What limits the capacity in a Kinesis data stream?

Answer 21

The number of shards. Each shard allows for 1MB/s incoming data and 2MB/s outgoing data.

Question 22

True or false: We can increase Kinesis Stream shard read throughput by adding more consumers

Answer 22

False. The read throughput is fixed at 2 MB/s per shard

Question 23

We want to modify data in real-time before sending it to the consumer? What service can we use?

Answer 23

Kinesis Firehouse. We can attach a Lambda to it to process data

Question 24

What is the partition key in a Kinesis Data Stream shard?

Answer 24

The partition key sperates data in the data stream into shards

Kinesis uses the partition key to determine which shard owns which data record