The y axis scaling is based on a functional decomposition of an application into different components. The y axis of Scale Cube represents the separation of responsibility by the role or type of data, or work performed by a certain component in a transaction. To split the responsibility, we need to split the components of the system as per their actions or roles performed. These roles might be based on large portions of a transaction or a very small one. Based on the size of the roles, we can scale these components. This splitting scheme is referred to as service or resource-oriented splits.
This very much resembles what we see in microservices. We split the entire application based on its roles or actions, and we scale individual microservice as per its role in the system. This resemblance is not accidental; it is the product of the design. So we can fairly say that y axis scaling is quite suitable for microservices.
Understanding y axis scaling is very significant for scaling a microservice architecture-based system. So, effectively, we are saying that microservices can be scaled by splitting them as per their roles and actions. Consider an order management system that is designed to, say, meet certain initial customer demand; for this, splitting the application into services such as customer service, order service, and payment service will work fine. However, if demand increases, you would need to review the existing system closely. You might discover the sub-components of an already existing service, which can very well be separated again since they are performing a very specific role in that service and the application as a whole. This revisit to design with respect to increased demand/load may trigger the re-splitting of the order service into a quote service, order processing service, order fulfillment service, and so on. Now, a quote service might need more computing power, so we might push more instances (identical copies behind it) when compared to other services.
This is a near real-world example of how we should scale microservices on the AFK Scale Cube's three-dimensional model. You can observe this kind of three-dimensional scalability and y axis scaling of services in some well-known microservice architectures that belong to the industry, such as Amazon, Netflix, and Spotify.