One of the most difficult tasks in an IT organization is to identify those applications which are no longer necessary to the business and wasting data center resources. For example, an application that is used very infrequently may appear to be idle, however it may be highly critical to the business for the short duration it is used. For the months it is sitting idle, should it be consuming server resources? How can an organization better understand the usage patterns of their applications to optimize the resources needed to keep them available?
Metering is the task of taking monitoring data (see Centralized Administration) about the PaaS instances and converting that data into knowledge about utilization. This metering data will show:
How much system resource did the instance consume?
What were the most frequently used applications?
Did the applications exceed any established resource quotas?
How many users were active on the applications?
This information can then be applied to a periodic billing event where the usage is converted into a financial amount that reflects a charge back.
Tenants of a PaaS solution are typically seeking a "pay as you go" cost structure to avoid large up front expenditures for hardware and software. This value proposition makes public PaaS implementations extremely attractive to smaller organizations looking to avoid building a data center.
But for organizations who already have a core competency in hosting their own applications, this same model can be effectively applied as a way to manage application usage. Tenants using a private PaaS solution can be charged back by the PaaS provider for only the data center usage their applications actually consume. While this may represent only internal funds within an organization, the key idea is that there is now accountability for what applications are being used and at attempt can be made at assessing their business value.