Deployment needs

[Deprecated] Rasa X Community Edition
1

Hi guys, I am writing here as I haven’t found so far any proper hardware recommendation based on one’s needs.

I know there’s a minimum requirements stated in the docs, but I am really interested in how many concurrent conversations that will handle. It’s pretty hard to plan for this as I have no experience in deploying an infrastructure that should be able to serve tens of people messaging on it at once.

I am interested in stuff like: For 10 to 50 concurrent conversations - what resources/deployment method works best

For 50 to 100 concurrent conversations - what resources/deployment method works best

For more than 100 concurrent conversations - what resources/deployment method works best

If anybody here has this info, I would really appreciate if you’d share it!

Thank you!

2

It depends on the frequency of a user sending the message. Each rasa-production container handles about ~20 messages per second. If you assume each user sends a message every 12 seconds or so, one node could probably handle ~200 concurrent users. So I would say the minimum requirement should sufficient for 50-100 concurrent conversations.