Both TCO and CPP analyses include inherent flaws that become acute when comparing two different types of technological answers to the same requirements. This is the case when comparing a monolithic chassis-based networking solution to a cloud-native, disaggregated one.

For example, today, in the IT and cloud domains, would any service provider or hyperscaler consider adding a new service to their portfolio based on a vendor specific, dedicated server when they have the option of a new virtual application running on high-end standard server? Even if the traditional single-service-server did present better CAPEX, can it really be compared in a TCO model to the immense OPEX benefits of the cloud shared resource model?

Technology Changed

Technology has changed. Datacenters have evolved into clouds. Today service providers can apply the benefits of the cloud to their own networks, and are starting to build networks like cloud, moving away from a vendor specific, single-service-router model to the model of an agile, shared network resource. The network is evolving from the traditional Physical Network Function (PNF) through Virtual & Cloud Network Functions (VNF/CNF) to a new model of Network Cloud Network Function (NCNF) which enables the full benefits of Network Cloud and with lower TCO.

TCO: A Flawed Analysis

TCO is flawed. It demands a tremendous amount of analysis effort in areas in which the differences between solutions are relatively minor. The fact that it involves many details also makes it very easy to skew.

More importantly, TCO analysis (as well as CPP analysis) neglects to measure solution functionality. That is, TCO only counts the total cost of an infrastructure (based on a given – i.e., capacity, architecture and port count) while neglecting to evaluate the yield from this infrastructure, per solution (e.g., functionality, capacity etc.)

These inherent flaws are acceptable when evaluating solutions based on the same technological and architectural concepts. However, when a new technology is introduced (such as DriveNets Network Cloud) and evaluated by these old tools, then it may lead to a wrong conclusion and significant losses for the service provider.