Under certain background experience, the final state can be directly considered
how to solve large flow, each try to deal with simple things, then load balancing:
- Access: BGP, DNS; (three load balancing based on IP)
- Tap into: core switches;
- Tap into load balancing: LVS; (four load balancing based on IP + Port, i.e., connection)
- Application-layer load balancing: Nginx / HAproxy; (seven load balancing, based on a request)
- Any Proxy: db proxy, cache proxy, app proxy and the like; (in stations or load balancing infrastructure, reducing the complexity of operations, the flow rate split, the computational load)
- Application Cluster: operational level load balancing, such as autonomous division, business logic split, split off-line, multi-level cache, a multi-stage hot standby;