The dimensions – the „sizing“ – of the server system is often oriented in these cases to the estimated or known peak loads for the application. The corresponding scaling behavior of this type of architecture is known as „scale-up“ because increasingly large resources must be made available in a server module for the application, i.e., it must become „higher and higher“.   

Pooling systems of this type require even larger and more powerful servers with the capability of managing more RAM and more CPUs and of making the resources available to the consolidated applications in an efficient manner. These high-end servers are more complex and expensive to operate and purchase by a factor of several times than standard servers. The ­decision for or against such monoliths is closely related to a possible continued use of the system currently in use, the residual costs which will be incurred, i. e. the point in time at which the exchange takes place as well as a sophisticated plan for the effective utilization concept of the newer system.

Grid computing allows flexibility   

In contrast to this model, modern, modular application architectures feature horizontal scaling, also known as „scale-out“. Horizontal scaling refers to the possibility of making parts or functions of applications available to more user queries by ­means of additional instances on an additional server and ­placing this system box, figuratively speaking, next to the first one. The ­encapsulation of business logic on one Web application server is one example of such an approach because it can as needed be launched as an additional instance on another server system.    

„Scale-out“ is a positive characteristic because it allows the use of small, standardized server systems, their merger into pools, and the allocation of resources according to demand. Surplus computer capacities can quickly be made available to other ­applications by shutting down superfluous instances of a particular application. Applications of this type and their environments can be consolidated easily because, in the extreme case, a single instance of the application is running concurrently on an available server somewhere.

The concept of modularity or layered architecture has been used for newly developed applications for years. But positive consolidation effects are realized from this only if a requirement profile is created which permits uniform utilization of the pool and does not require large overcapacities to cover common peak loads by the interaction within the application landscape and the pool utilization.