BLB provides load balancing services through a cluster. When a server in a load balancing system cluster fails, other servers will automatically take over the work of the failed server to ensure high availability of services. The basic architecture of BLB is as follows:

The processing flow of an access request is introduced taking TCP and HTTP listeners as examples:

1.The user sends a request to initiate a TCP handshake request to the service's listening address and port; 2.In 4-layer load balancing, find the addresses of all backend cloud servers configured by the user, then select a currently available cloud server based on the configured load balancing policy (polling or least connection first) and initiate a TCP handshake request. 3.If the user initiates a request for 7-layer load balancing, the 4-layer load balancing in Step 2 will be directly transferred to the 7-layer load balancing device, which will initiate a TCP handshake request to the backend cloud server. In addition to fully supporting 4-layer in terms of features, the 7-layer load balancing also comes with features such as real IP and keep session. 4.After the connection is successfully established, the mapping transformation for the connection is completed, and the frontend request is forwarded to each cloud server. And during the entire work period of BLB, health checks are performed on each cloud server according to the configuration, and unavailable cloud servers are blocked.