How does load balancing calculate the number of visits to all LAN IPs under the same public network address?
The load we are currently using is a product of a certain manufacturer (to avoid suspicion of advertising, the name of the manufacturer will not be mentioned). The load provides a public network address, and then the whole country accesses this address, and the load distributes it to each node. However, during testing, we found that if We have 50 people in an office in Beijing, and the export is the same public network IP. In terms of load, all access from this public network IP is uniformly distributed to the node to which the IP manual is distributed. Can this be distinguished in terms of load (according to The traffic coming from the public IP address is distinguished by each LAN address and distributed to different nodes)?
When we visit an office at the same time, we check the number of connections on the node visited for the first time. The number of connections will be added according to the number of connections. I don’t know what parameters can be adjusted on hard load or soft load such as NGINX?
滿天的星座2017-05-16 17:26:40
It depends on how many layers the load balancer you use supports. If it is layer 7, it can be differentiated into applications, which means that each application layer protocol can be differentiated. Even if it comes from the same public IP, it can be differentiated through different Depending on the session ID, you can configure the load strategy (load according to the session, which can be distributed to different nodes). If layer 4 is supported, differentiation can also be achieved through different ports. If it's 3rd floor, there's nothing you can do. Contact the manufacturer directly for after-sales technical support.