Cloud Computing's capital spending vs. operating expenses

The debate about the economic benefits of cloud computing is so intense that it is often attributed to the controversy between operating expenses (OPEX) and capital expenditure (CAPEX). However, as with many debate topics, the conflict between the two is actually a cover horse, which masks the real source of conflict.

In the debate over capital expenditure vs. operating expenses, the fundamental question discussed is the debate about the future of IT infrastructure and operations teams: will they be the operators of the assets owned by the enterprise, or will they be the operators of assets owned by external suppliers?

It's really hard to understand: why spend hundreds of millions of of dollars to buy assets, assume responsibility for running them, and deploy the appropriate human resources (technology assessment, supplier relationships, capacity planning, etc.) that are considered more attractive than the assets that are directly provided by external suppliers (responsible for all operational responsibilities)? Another possible nightmare is that infrastructure and operations teams will no longer be involved in cloud computing operations, and that the application team will be responsible for direct communication with external suppliers, and that infrastructure and operations teams continue to be responsible for the inherent assets of the enterprise that is constantly engulfed by cloud computing.

There are heated discussions about the cost of cloud computing, mainly around capital spending and operating expenses, but most discussions around this topic do not fully understand the far-reaching implications of different funding models and their mapping of the future of IT applications.

In my opinion, these discussions are all about determining the best way to run an application, because the application is all the value it creates.

The following are the thoughts and implications of the current financial model that is not understood in the discussions on capital expenditure and operating expenses:

Operating expenses should be more expensive than capital expenditure

One of the benefits of operating expenditure patterns is that there is no long-term constraint. Once the user has run out of resources, the resource can be returned to the supplier, the supplier has ownership of the resource, that is to say, the supplier needs to find out how to make full use of the resources to guarantee its economic benefits.

The absence of long-term constraints is certainly cost-effective, as users do not need to make huge long-term investments. So it's reasonable to say that from every unit of measure, operating expenses should be more expensive than capital spending. We can look at the price of car rental, in fact, we are only for short-term use of resources to pay an additional fee. Since it is a short-term use, renting a car is certainly more cost-effective than paying a large price for a car.

How to calculate

Therefore, the real question is not which option is to pay more for each unit of measurement, but which option is more expensive for the total amount of resources (or the total amount of all resources used by the operating application), which is trickier than the cost per unit calculation. First, it needs to predict the total usage for a certain period of time (usually one months). In other words, how many hours does the application run per month? How much storage does it use? How much network traffic will there be?

Second, this calculation may have to consider changing the lease rate based on the use layer. If the total storage uses 10,000 megabytes, the gigabit storage is at a certain rate level, and it will be cheaper if total storage uses 10 megabytes.

Third, this calculation also needs to consider the different use patterns, some periods of low usage and other times of high utilization rate. For example, a financial services company's application may have a monthly peak, a quarterly peak and an annual peak, not to mention unpredictable peaks affected by aperiodic events, such as amending the law----Changing the attractiveness of certain types of financial products to consumers.

This tricky calculation needs to be compared to the TCO, O represents operations, not ownership. For many usage patterns, it is possible that leasing patterns are more attractive than cumulative usage patterns, even more expensive per unit of measurement. Take the car rental for example, even if the price of a car rental is more expensive than the purchase of cars, if one months only five days to use the car, then the car rental must be cheaper.

In other words, economic assessment should be more complex, given the different uses and the different costs associated with the asset leasing model. Of course, regardless of the usage pattern, the cost of the asset ownership pattern is the same, whether it is used daily or not more than one hour per month.

Other resistance factors affecting operating expenses and capital expenditure

Of course, there are other factors that affect this calculation. Renting a car will charge you according to time, and you will need to take the leasing process, listen to the extra insurance, and check the car lights. If a person needs to use a car for one months and eight days, a car is a good deal. But sometimes people may choose to buy a car after hearing it repeatedly, "How do you plan to deal with the insurance of the car?" Our top service allows you to not have to pay for these fees.

In other words, these resistance factors may make you change your choice. Ronald Coase, the Nobel laureate in economics, described the cost of car leasing as "transaction costs": the costs associated with economic transactions add an additional burden to this option.

In the case of IT applications, there are often more such factors, often allowing organizations to try to avoid touching them once they have started the application. For leasing models, the "simply set aside" approach is not a good idea from an economic standpoint.

In leasing mode, when users want to assess the use of assets, it is clear that they should be minimized. James Staten, cloud analyst at the renowned Forrester Institute, describes this approach as "meeting the minimum Requirements": people should try to minimize the resources of the application as long as their functional and performance requirements are met. If you don't use the requirements, the application should shut down and open again when you need it again.

However, there are so many resistance factors involved in resource management that most IT teams in the discussion of operating expenses and capital expenditures assume the full use of the highest amount of resources required for the application's peak load, and then use this as a basis for evaluating asset leasing patterns and asset ownership patterns. This calculation is simply unreasonable. In fact, most companies do not even consider other options, because choosing a different mode of operation does not eliminate these friction factors.

Capital expenditure vs the future of operational spending

In a properly managed cloud environment, this drag factor is decreasing. We can use tools to automate the instantiation and termination of resources, and even the application itself can automate the management of this process without human intervention. In this environment, "Transaction cost" is more feasible than "Meet the minimum demand".

Overall, lowering operating transaction costs means that the application operations model needs to shift over time. Conversely, this means that financial analysis will need to change as it enterprises begin to reassess application resource consumption patterns. Many application designs will shift to a certain level of continuous operation of resources, as well as additional and reduced resources for changing usage patterns. The end result might turn this critical point calculation into an asset-management model, not an asset-ownership model.

At the very least, this avoids the assumption that the full time peak load resource operations are calculated, while incorrectly identifying asset ownership patterns is the most feasible approach.