Forums

Hello forums!

An event, for which I am designing a registration portal, will be opening registrations soon and we need to serve approx. 300 users at once, for a duration of about a half an hour. I have been trying to find some information on this topic, but I only found some fragments.

How to ensure that? I know, that the amount of requests I am able to serve per second is the number of web workers times the time of one request.

The request time is the time written in access logs? Is that in seconds, or in miliseconds? How many web workers to buy, if my account has 5 running web apps and this is only the case in one of them? I am also considering creating a new python anywhere account for this project.

Is there then a way to test this resilience, or does it simply have to work on first go with real participants?

Thanks for all the thoughts.

If this website will have very different performance requirements to your other websites, I'd definitely recommend putting it into a separate account so that you can get more worker processes for it without having to pay for an increase in the workers for your other sites.

The time taken to handle a request is indeed the one in the access logs, and it's in seconds, so response-time=0.004 would mean that your site took 4ms to respond to the request in question.

For testing -- that's a bit trickier. We use Selenium and Playwright heavily to test our own systems, so perhaps you could put together a load test using one of those? I would recommend Playwright these days, it's easier to use than Selenium.