30 likes
·
7.3K reads
2 comments
SNS definitely supports batching as of 2021.
SNS also has a very robust re-try policies, including configurable HTTP endpoint re-try policies. For example, SNS will keep re-trying to deliver messages towards other AWS endpoints such as SQS or Lambda over 23 days with 100,015 re-tries. Note that 23 days is a lot longer than SQS' own default retention period of 4 days.
Just curious, it seems like you'd use a fanout strategy so that a you can persist messages and retry in case pushing the message from SNS fails?
Is that correct?
Are there other reasons to use a fanout strategy with multiple SQS queues subscribing to 1 topic, instead of subscribing the actual compute source to the sns topic (lambda, api gateway, etc.) ? i.e. cutting out the SQS queues (middle man)