What algorithm does Nginx use to load balance? Can it balance based on connection load?
Currently, Nginx uses a simple round robin algorithm. It cannot balance based on connection load. (This may change in a future release.) There is a third-party module, Http Upstream Hash, that uses a form of hash load balancing. Note: Many users have requested that Nginx implement a feature in the load balancer to limit the number of requests per backend (usually to one). While support for this is planned, it’s worth mentioning that demand for this feature is rooted in misbehaviour on the part of the application being proxied to (Ruby on Rails seems to be one example). This is not an Nginx issue. In an ideal world, this particular problem fix request would be directed toward the backend application and its ability handle simultaneous requests.
