Load Balancing is an important technique in cloud-native application design to achieve scalability, reliability, and availability. The load can be distributed among nodes (physical or containers), based on rules like round robin, weighted, performance-based, geographical distribution, etc.
Load Balancing can be achieved at the following levels
DNS Level: DNS level load balancing is a method of distributing incoming network traffic across multiple servers or IP addresses by using DNS (Domain Name System) servers to resolve domain names to IP addresses. You can choose distribution riles based on need, for example, you might want to send traffic originating from Europe to hit Europe servers whereas traffic from North America to hit North America servers. While resolving the DNS, the traffic manager will choose the backend endpoint based on the rules set.
Layer 7 or Application Layer: In Layer 7 load balancing, the load balancer analyzes the content of the incoming requests, including the HTTP headers, URLs, and other application-specific data, to determine how to distribute the traffic. For example, we can set rules that /images pattern is getting redirected to a backend, whereas /videos pattern is to another. Additionally one can have features like SSL termination, and WAF (Web Application Firewall, that will protect from threats like SQL injection attacks, Cross Site Scripting or XSS attacks, etc.) implemented.
Layer 4 or Transport Layer: Layer 4 load balancers can route traffic based on basic criteria such as source IP address, destination IP address, source port, destination port, and protocol type. At the transport layer, the load balancer does not have access to request data, hence decisions can only be taken at IP or Port level. At the same time as no parsing is involved, the overall performance is better.
Related: https://kamalmeet.com/cloud-computing/azure-load-balancing-options/