load balancing

  • Round robin
    • load distributed sequentially & evenly across available servers.
    • Eg:
      upstream backend {
        server app1 : 3000;
        server app2 : 3000;
        }
      
  • weighted load balancing
    • you weight servers. % wise
    • Eg:
      upstream backend {
        server app1 : 3000
        weight = 3;
        server app2 : 300
        weight = 1;
        }
      
    • app 1 get 75% more traffic.
  • least connections
    • traffic to least busy instance.
    • Eg:
      upstream backend {
        least_conn;
        server app1 : 3000;
        server app2 : 3000;
        }
      
  • Health check/failover
    • if app dies, traffic moves to app2.
    • Eg:
      upstream backend {
        server app1 : 3000
        max_fails = 3
        fail_timeout = 30s
        server app2 : 3000;
        }
      
  • sticky session load balancing
    upstream backend {
        ip_hashserver ...
        }
    upstream backend {
        sticky cookie srv_id expires = 1h;
        server apps ...
        }
    
    • can be based on IP or cookies etc.
    • not a load balancing algo, but rather a constraint that modifies load balancing behaviour.

This site uses Just the Docs, a documentation theme for Jekyll.