#cloud school

This blog is part of Cloud School series of blogs

Load balancers are quite common in architectural designs. Load balancers are primarily used for spreading requests across a pool of servers. They are also called reverse proxies. Note that they are different from forward proxy (like squid). Forward proxy sits close to the client and reverse proxy sits close to the server.

Load balancers have the following features

Persistence - Load balancers can maintain session persistence which allows the requests from same session to go to the same backend server.

Health Checks - Load balancers can periodically check health (and adjust weights) of the backend servers. This ensures that service remains up even when a few backend servers are down.

Metrics - Since all the requests and responses pass through load balancers, they are often used for collecting the service metrics

SSL offloading - Load balancers often terminate HTTPS/SSL and store client facing certificates. Connection from load balancer to backend can be encrypted again often using a low CPU intensive cipher like AES-128.

Request Queuing - An often overlooked but great feature of load balancer is to queue the requests instead of overload the backend servers. This improves the service response time and can handle small traffic spikes without affecting service health.

DOS protection - Load balancers can maintain metrics on requests per ip per min etc and can throttle/deny abusers.