Hacker News new | past | comments | ask | show | jobs | submit login

> Broker instances start at $0.21/hr and broker storage is $0.10 per GB-month

Kafka requires Zookeeper, do you pay for zookeeper instances too? Anyway, this pricing seems prohibitively costly unless you really need hands-off approach to your infra. Kafka is pretty low-maintenance and stable in my experience.




From the pricing page: https://aws.amazon.com/msk/pricing/

"You do not pay for Apache Zookeeper nodes that Amazon MSK provisions for you, or data transfer that occurs between brokers and nodes within clusters."


Hey alienreborn,

sorry that I answer here, this is totally unrelated to the topic being discussed. I've just seen that you replied me on an old HN post of mine and cannot reply there nor can I find another way to contact you on your profile.

It's regarding the habit tracker https://everyday.app I'm working on. You wrote "Your web-app is the best habit tracker ever created and I used it for a while after I paid for it very early on. :) However, I left it because mobile app is not as smooth of an experience. Currently using habitify.me across my devices but will switch back once mobile app catches up with other apps." and I'd totally love to hear in more detail what was lacking or what do you think I should get fixed to help the app catch up. Feel free to answer here or send me an email at joan@everyday.app.

Sorry for reaching out this way and thank you! :)


Just to note, the $.21/hr broker is on an m5.large (2 cpu, 8 GB mem), which goes for $.096/hr.

We run three nodes right now on m5.xlarge instances. At $.42/hr for the managed kafka, compared to $.192/hr self hosted kafka, I think we'll keep it self hosted for now...


What about the cost of the ZK nodes and associated EBS?


I highly recommend https://aiven.io/ which has lots of managed services including Kafka. Their pricing might be better at low-end.


I find that this is true, but when it breaks, man it breaks REALLY hard.


It does. And until last year even the major support company's containers (for Kafka etc) were not very good, we didn't upgrade a k8s cluster purely because of the state of their stack. This stuff scared me and this is a shocking affair these days. In some ways it's a step back in reliability, and i'd have taken this AWS option in a heartbeat, price-gouging aside.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: