According to Cloud Monitor 2021, 97 percent of all companies are already in the cloud or are planning to get started; cloud services are still not an issue for only three percent. Even if the share of IT costs for cloud computing has not changed at 20 percent compared to the last survey, more and more companies are relying – already now or in the future – on more complex cloud structures consisting of different providers and models. In this context, cloud cost management is becoming central. Because the more complex the cloud structures are, the greater the risk of producing unnecessary costs or even losing track of the total costs. Claims can quickly cause six-digit burdens.
IT on-demand: Billing models for cloud services
The first and most important step for successful cloud costs optimization is a strategic and holistic introduction of cloud solutions. Those responsible should determine all relevant questions and financial aspects in advance. A central key point for efficient cost management is the choice of the optimal billing model. Because cloud services are calculated very differently and depending on the respective service, provider and usage. There are the following three basic variants:
1. “Pay As You Use” / “Pay As You Go”
This is the most widely used billing variant, in which the costs are linked to the use. The diverse characteristics can in turn be divided into two variants based on their service models:
- Software as a service (SaaS)
The provider provides software via cloud and typically calculates the use with a license price per user. These costs per user license usually decrease with increasing number of users, but increase when the company has special requirements – for example, higher availability, conditional access or faster response times.
- Infrastructure (IaaS) and Platform as a service (PaaS)
Here, the provider provides an infrastructure or platform that the company uses per unit actually used (e.g. minute or function call). The cost of such cloud instances results from parameters of their configuration (e.g. CPU performance, memory, cache memory, database type) as well as the utilization of resources (e.g. computing time, exhausted storage space, incoming and outgoing traffic).
2. Reserved instances
Companies reserve fixed instances for a pre-defined period of time and receive a discount of up to 60 percent in some cases. A reserved instance is a kind of coupon that can also be used for other machines (e.g. in the event of a failure) or other tasks of the customer. Reserved instances must always be paid by companies – even if they do not use them.
3. Spot instances
With this not very common billing model, providers offer unused capacities with discounts of 70 to 80 percent, but reserve the right to terminate them at any time within one to two minutes. Typical use cases are non-time-critical applications such as batch processing, big data workflows or complex analyses (AI/ML), which can be interrupted and restarted as desired.
If you are looking for a good solution, consider devops as a services by itoutpost.
Further adjustment screws for cloud managers
In order to opt for a billing model, those responsible already have to clarify numerous aspects. In addition, there are other questions that you should consider from the beginning:
How does the migration take place? Is “Lift-Shift” sufficient – i.e. a shift with minimal changes – or does another method of cloud migration possibly allow even greater savings? How is the data backup or archiving? How can the purpose and storage class be optimally linked to minimize costs? Who is responsible for monitoring? Which data and costs are checked and in which cycle? Which deviations from the target are tolerable and from which point are adjustments necessary? What adjustments are these?
If those responsible plan the cloud holistically and also think about cost management from the beginning, they have at best a framework that informs them in good time about deviations and specifies further procedures. The crucial thing is that those responsible should never underestimate cloud costs and successful cost management as a success factor.