Operations are tricky, but in my experience it was plenty fast

Are you sure you need all the functionality that onyx provide? Using onyx can be worth it for complex stream processing , but for simple use cases it is overkill IMO

So what is the delay using onyx vs do operations immediately on API request?

I am not sure 🙂

That is why I am here.

I am thinking about onyx not as solution for complex processing, but as solution to scale number of workers to convert data to PDF.

But maybe better idea is to run more instances of service in kubernetes.

I don’t feel confidence about scaling up

steps for precessing are probably simple: 1) fetch data every X minutes / manually by UI / by API request 2) use this data with template to generate PDF 3) return / send / save PDF

I don't thinks onyx suits your use case

How are you running this? on AWS/GCP/Azure?

I will use Google Cloud (at least this is a plan)

I'd either put the fetched data on a queue (RabbitMQ, or something similar) and create workers that read from this queue, generate the pdf and save / send it

this way you can add more worker nodes if you need more capacity

Could be why queue system, but not onyx?

It's easier to setup and operate

With onyx you will need to run a zookeeper cluster, a likely some queue as well to get your data into onyx

Oh I though onyx is more independent solution

Why there is no work about that in readme?

so it looks like onyx environment is maybe unnecessary complex for me

thanks for that info