@yashovardhan as per our chat on telegram you’ve suggested it can handle around 20 keys per second. We need to know more about this:
1, Do you throttle api requests?
2, If not what’s the impact on performance for a larger rps.?
3, Is there anyway we can get this number increased?
4, Do you have a document containing the metrics about performance?
Our expected use case will have spiked traffic where the spike can have upto 1000 requests per second to you.
@girichaitanya11 Just to clarify the system load depends on various factors like location etc and the load we are talking about in this question is only specific to new user registrations. It can scale up to 80 new user registrations, whereas logins for existing users are not restricted and can be done in 1000s per second.
The reason behind this is we use a distributed key-generating mechanism for new user registration (key generation) and it uses a consensus algorithm that needs consensus to generate these keys, we are working on the next iteration of our distributed key generation tech which significantly improves these numbers of registrations.
As of now, there are various mechanisms to overcome this, some of them are:-
Pre-registering users: If you know the user’s id or email ahead of time then you can register the user prior to the user’s actual login and since once the user is registered system is capable of handing 1000’s of req per second it will work without any issue.
Using a queue:- If you don’t have the above info ahead of time you can create a queue in your login backend for registered users.
Our team can help your org with both of the solutions or can come up with some custom solutions based on your authentication flows.