Performance best practice documentation for large scale server managers

[moranno]

Currently, I am using an older version of Shadowsocks (non-ss2022) to build my service, and it is encountering severe performance issues (CPU usage often exceeds 150%) when there are a large number of users (over 10,000 available users and 200+ online).
Are there any strategies or methods for performance optimization? I would appreciate any suggestions you may have.
Thank you!

[maddyhof]

Labeling this as a feature request for creating some best practice documentation for large scale server managers.

In the meantime, having multiple servers and a dynamic key service that routes users to servers based on capacity would probably be a decent move.