An ELB instance receives inbound traffic from clients and routes requests to real servers in one or multiple AZs for processing.
The load balancing service is mainly provided by an ELB listens on the ELB instance for requests and excecutes corresponding policies to distribute requests to services such as real servers. you can configure the forwarding protocol and protocol port in client-ELB and ELB-real server dimensions to have ELB directly forward requests to real servers.
You are recommended to configure backend ECC instances in multiple AZs for an ELB instance. In this way, if one AZ becomes unavailable, the ELB instance will route the traffic to other normal instances in other AZs so as to avoid service interruption caused by AZ failure.
A client request accesses the service through a domain name. Before a request is sent to an ELB instance, the DNS server will resolve the ELB domain name and return to the client the ELB IP address that receives the request. When the ELB listener requests, it will use different ELB algorithms to distribute them to real servers. currently, tencent cloud suppotrs multiple load balancing algorithms such as WRR, ip-hash, and weighted least-connection scheduling.
An ELB instance can monitor the running status of real servers to ensure that traffic is routed to only healthy servers. when the ELB instance detects tha a real server is exceptional, it will stop routing traffic to the server and resume routing after detecting that the server runs normally again.