AWS SQS, RabbitMQ, or Redis. I would build on one of those 3 first, then migrate when throughput became an issue. I've had good experiences with Kinesis Firehose as well.
I've yet to run into a workload that actually needed Kafka (and if you're not a Java shop don't add Java pieces unless you know you need what those Java pieces provide).
I'm amused by the amount of third parties (Sumologic for example) that have ServiceNow plugins. I'm curious what kind of usage metrics they have on the feature. I suspect it's just a feature parity/selling point but not actually used in practice, like EKS and managed SFTP on AWS.
Nowhere I've worked that used ServiceNow encouraged solving practical problems with it. A grand total of once I took a crack at something that would have been trivial if Sumologic could alert to ServiceNow. It became hilariously obvious in 3 or 4 back and forth emails with the gatekeepers that their 3 month testing process, 45 emails, and 16 meetings requirement for a one line change was not cost effective.
My time was better spent crafting an email forwarding rule that sent the related conversations to the trash and moving on to solvable problems.