Are you also looking for the cheapest cloud instance? Don't worry: you are not alone. Google returns 2.2 million hits when you search for the “cheapest cloud instance”. Even if you were able to search through all the results, you would be none the wiser. Because cloud costs also rely on the old consulting maxim: “It depends...”
For a long time, the issue of price was not open to discussion when using the public cloud. A glance at price lists that show 0.9 cents per CPU hour prevented the public cloud TCO from being questioned. And of course there was the killer argument: the costs are based on the actual use or, even better, on the business actually supported. Together with auto-scaling, which automatically adjusts resources to the current workload, it's an absolutely fantastic thing! But often only in theory. The monthly bills that large companies using public clouds receive at the end of the month are said to have brought many a business executive to the brink of a heart attack. There are two main reasons for high cloud bills: wasted cloud resources and pay-as-you-use payment models
In its "State of the Cloud 2019" report, Rightscale has, among other things, shown the percentage of wasted cloud resources according to the users. The result (which is probably more a gut feeling) is: 27 percent. This means that about one quarter of the paid costs for using the public cloud are for nothing. A generous tip for the public cloud providers! Other estimates assume even higher percentages. That may or may not be true. However, it is undisputed that cloud resources are not aligned with actual usage in real life. And that there is potential to optimize the costs for the public cloud. This becomes even more relevant when companies develop a multi-cloud strategy.
But cost control is no trivial matter. The price structures and scope of services vary dramatically among the various providers. Differences already start to emerge when defining a CPU hour. Network costs, storage performance, management services – the complexity of cloud solution packages makes pricing even more confusing. This is the external factor that makes it difficult to compare costs directly.
The other factor originates from the cloud users. There are legendary stories where developers forgot to "turn off the lights" – in other words, to switch off extensive cloud resources before the weekend. While the developer was enjoying his weekend, the cloud resources were producing costs without providing any real value for the company.
The first step towards cost optimization is cost transparency. In a multi-cloud world, you should first of all think about how you can automatically gain transparency over the use of different clouds. You may also include your private clouds in this usage assessment. You will certainly be surprised at the potential for cost reductions.
And if you want to tackle the topic of cost optimization, remember that no two CPU hours are the same. Over-commitments on the part of public cloud providers can significantly reduce the actual capacity delivered. A better decision criterion than the list price alone is the real price/performance ratio, which can vary considerably despite formally identical performance (“on paper”). This can be determined by means of benchmarks. Using each public cloud according to its (technical) strengths is one of the key success factors for a multi-cloud-strategy. If you are able to control the cost of resource utilization in your multi-cloud universe, you are in the best possible position.