The cloud is there – but the money is gone?
An extensive assessment has confirmed the added value afforded by the cloud to the company and finds the costs justifiable – in short, the business case is a positive one.
An initial PoC also confirms the results, and so the cross-company, or rather cross-application, launch begins.
One system after the other is migrated to the cloud, and, apart from a few smaller frictions to be expected, everything goes as planned. But then comes a rude awakening with an unpleasant phone call by the CFO, who just noticed that the company's credit card has been completely maxed.
At first, the answer is simple: the costs have gone through the roof. But on closer inspection, the difficult question of the "Why" remains unanswered.
The many possibilities and freedoms that come with the cloud always carry the risk of unplanned and often unwanted costs. In contrast to operation in your own data center, in the cloud environment every decision, every change to the planned architecture and every additional resource entails new, additional costs. The time of "business-as-usual" costs is over.
Well-thought-out cost management is thus extremely important for the transparent and comprehensible listing of expenses. But at the same time, you don't want to lose the advantages of the cloud in terms of speed.
At this point, you can and should avail yourself of the governance tools and services of the cloud platform, such as Azure Cost Management and Azure Policy.
Azure Cost Management helps you depict and analyze the costs. Through a separate set of rules, Azure Policy will help you enforce the necessary transparency by means of tags, while various restriction options will prevent cost explosion.
But doesn't the cloud offer many cost benefits as well?
A legitimate and often-mentioned argument for the cloud transformation is cost saving. Being able to adapt your company to current conditions is important, and not only in times of crisis. This holds not only for the business model, but for IT as well.
It must be possible to adapt peak loads both downwards and upwards for dynamic and attractive pricing.
Below we explain in more detail the extent to which the cloud can support you in this regard.
CapEx vs. OpEx – old hat
This topic is not new – and in keeping with the slogan "Avoid high investments and fixed costs".
For many years, arguments like this one have appeared when pleading for hosting or outsourcing in general. It continues to be a current issue regarding the cloud as well. But what has changed?
On the one hand, IT outsourcing has increased significantly and is still growing, so that we should be speaking of OpEx versus OpEx. On the other hand, the definition of OpEx in the cloud context is changing towards use-based provision, making OpEx vs. OpEx (cloud) an interesting question.
5 ounces of cloud, please
You can buy only what you really need as a private person at the butcher's, and maybe as an entrepreneur as well. But when it comes to IT services, things get more complex.
Irrespective of whether you're operating the largest share of your IT at your own data center or outsourcing it to an external service provider, in the end you always face three central questions:
- What volumes with what features do I need at the moment?
- What and how much will I need in the future?
- How can I make sure that the costs for normal operation are appropriate without IT becoming the limiting factor?
There are two approaches to answering these questions:
On the one hand you can plan your own current and future requirements and aim at considering even the smallest factor in your planning.
On the other hand, there is the possibility of using a model having the necessary flexibility for responding to any future changes without causing too much waste, i.e. idle time costs.
While detailed planning certainly has its benefits, it often replaces chance by error, as Einstein once said.
Since even the best planning can't consider all possible influencing factors, a combination of both approaches is recommended. A cloud solution can especially afford great benefits here.
Flexibility and reserved instances – a contradiction in terms?
The biggest cost factor of an IT system usually is the VM, i.e. computing power in the form of CPU and RAM. There are two different models for utilizing these services in the cloud.
The first model is use-based payment, better known as pay-as-you-go.
In this model you pay for the booked computing power on an hourly basis according to actual usage. Pay-as-you-go is a suitable scenario for systems used only for selected tasks and remaining switched off for the rest of the time.
This model is highly recommended especially at the beginning of the cloud transformation, since the actually required resources may deviate due to hardware exchanges. Switching to a steadier model is recommended only once you know which resources you actually need – the same applies to new systems or project systems.
The second model allows reserving computing power for a longer time period, that is, booking so-called "reserved instances".
This model affords planning reliability to hyperscalers for the capacity utilization of their resources, which in turn come at an attractive price.
You can book reserved instances for a period of one or three years with monthly payments. Usually, most reservations are placed for productive systems that can be used and operated 24/7.
Even if it is possible to change or even cancel an existing reservation under certain conditions, this model involves a loss of flexibility.
It must be considered whether the lower price can compensate for the loss of flexibility; the more resources are operated the lower this risk becomes. The probability will then increase that you'll be able to employ a reservation for other purposes.
In general, deciding between costs and technical flexibility is always a personal matter and should depend on your overall strategy.
The price/performance ratio must be right, and the costs should be transparent and kept under control.