update entry on SQS

zk/Messaging_services.md

@@ -4,21 +4,6 @@ tags: [AWS]
 
 # AWS Messaging services
 
-## SQS
-
-> SQS: Simple Queue Service
-
-SQS is a service that allows you to send, store and receive messages between
-apps and software components built in AWS, with automatic encryption. It helps
-with decoupling and scaling.
-
-As the name indicates, its operating mode is that of a
-[queue](Queue.md) data structure offering first-in, first-out
-and other queue implementations.
-
-An example application of this would be to set up an SQS queue that receives
-messages and triggers a lambda whenever a new message is added.
-
 ## SNS
 
 > SNS: Simple Notification Service
@@ -30,3 +15,5 @@ through SMS, text, push notifications and email.
 
 **SNS pushes messages out to subscribers while SQS stores the messages until
 someone reads them.**
+
+Relation between SNS and SQS - typically paired aren't they.

zk/SQS.md

@@ -0,0 +1,52 @@
+---
+tags: [AWS, SQS]
+---
+
+# AWS SQS
+
+SQS ("Simple Queue Service") is a service that allows you to send, store and
+receive messages between apps and software components built in AWS. It helps
+with decoupling and scaling.
+
+> Amazon SQS is a distributed queue system that enables web service applications
+> to quickly and reliably queue messages that one component in the application
+> generates to be consumed by another component where a queue is a temporary
+> repository for messages that are awaiting processing.
+
+As the name indicates, its operating mode is that of a [queue](Queue.md) data
+structure offering first-in, first-out and other queue implementations.
+
+## Example use case
+
+![SQS example diagram](../img/SQS-example.png)
+
+A request is made to an [API Gateway](API_Gateway.md) endpoint with a body. The
+body is then parsed and inserted into a database.
+
+The benefit of adding SQS as a buffer between the request and the updating of
+the database:
+
+- It can better handle spikes in activity, buffering requests to the database
+  until it is ready to handle them. This prevents the messages getting lost if
+  the server is overloaded
+
+- There is a retry mechanism built into SQS. If the database write fails, the
+  message stays in the queue allowing for retries
+
+- It facilitates decoupling. Without SQS as the middleman the responsibilites of
+  the lambda would be compounded - it would receive requests and update the DB,
+  plus any additional processes such sending a message to SNS. In the solution
+  we have two lambdas co-ordinating actions in a decouped manner via SQS.
+
+## Dead letters
+
+As SQS allows for multiple retries we could end up in a situation where a
+malformed message is continually processed in a loop. To avoid this you can set
+a maxiumum retry limit and, when this is exceeded, shift the problematic message
+to the dead letter queue (DLQ) and remove it from the main queue.
+
+![SQS deadletter example](../img/SQS-deadletter.png)
+
+If the DLQ reaches a certain threshold this can trigger additional handling such
+as raising an Alert in [CloudWatch](./AWS_CloudWatch.md) or other monitoring
+tool.