Google Vision API. How to distribute the load?

Question

Google Cloud Platform

Suppose you want to determine what is shown in the photos that users upload through the client. The algorithm is as follows: the client sends a request to the application API with a photo upload -> the server accesses the Google Vision API -> receives a response and writes it to the database. The response from the Vision API takes 2-3 seconds. Every second, 5,000,000 such users need to be served. Is it enough to connect more instance servers to withstand such a load, or is the task still impossible? Maybe there are other ideas that I just don't know about? Thanks to.

PS: The application is written in PHP

Answer 1

what's the question?
up - min (API limits, channel-server-perc)
to you - max(channel-server-perc)
set the server as the combination of these indicators
grows with anlim servers - the answer is "enough"