Load Balancing is a technique used to aggregate multiple ISP connections to maximise throughput and resilience through efficient distribution of internet traffic. Should any single connection fail, traffic will be re-routed down the other available internet connections to ensure continuous uninterrupted connectivity.
Outbound Load Balancing
During peak traffic periods, the network can become congested due to certain applications consuming too much bandwidth. Given the low bandwidth capacity, the network will queue up data packets or trigger data packet loss. Load Balancing provides multiple Internet connections into an organisation's network instead of a single connection, which can be grouped and logically controlled.
All available bandwidth is optimised by distributing traffic through each active Internet connection according to the distribution algorithm chosen. This ensures data packets from the original stream are not forwarded out of order.
Inbound Load-Balancing (DNS)
Inbound traffic requests from external users into the organisations network are handled by the Round-Robin algorithm, which intelligently distributes external requests over multiple Internet connections, instead of one. When an external request for the domain name of the web server is received on WAN 1, the domain name request is passed to the DNS server within the appliance. The DNS server then prepares a suitable IP address according to the Load Balancing Algorithm, which checks the status and load for all active Internet connections.
In diagram 1.2 the Load-Balancing algorithm for example, when configured for weighted least traffic determines that WAN 2 has the lowest load and least number of sessions, therefore instructs the DNS server to use that specific IP address. A reply is forwarded through WAN 1 to the original DNS requests source IP address. The source address receives the reply and retrieves the information it requests through WAN2. The web server is now accessed through the public IP bound to the Q-Balancers WAN port.
