We have seen the use of Compute Engine as IaaS, including Kubernetes clusters, which also rely on Compute Engine VM instances, and persistent disks. We don't have to worry about issues such as the proper distribution of internet traffic, requests, or processing when we use only one VM instance. Even in the case of overload, we can always scale our instances vertically and fulfill the processing requirements. But when we use more than one VM for a heavily accessed application or multi-tier application, load balancing becomes essential.

We may have to guide the traffic to the machines serving the corresponding tier, or we may have to guide it towards the same tier, while making sure that none of our machines get overloaded while other machines are chilling out. Fortunately, GCP provisions quite a sophisticated load balancer, which allows the routing of internet traffic based on the requested instance group (tier in practical application), available computing, and storage capacity, as well as distance from the region of the instance with respect to the user. In this chapter, we will go through HTTP, TCP, and network load balancing with reference to their concepts and implementation.

This chapter covers the following topics:

• The importance of load balancing
• HTTP(S) load balancing
• Other load-balancing rules