The scheduling algorithm is a wide category of strategies based on some statistical criteria. The idea here is to be more accurate about the way load distribution is done regarding the available resources in the backend servers. The most common criteria are as follows:
- By request: The distribution is based on the number of requests served by the node. Note that each server node can have a weight associated with this distribution to bias the distribution if a machine is less powerful than the other.
- By traffic: This is the same sort of distribution as the previous one, but instead of counting the requests, it uses the transmitted bytes.
- By busyness: This is also based on the number of requests, but only the live number. ...
