Cloud pricing models vary dramatically. Elastic utilization can mean wide variability in month to month. Make sure your financial goals like flat spending or opex versus capex match up with your cloud providers pricing model.
One of the benefits of cloud computing that’s often touted by providers is cutting costs: rather than having the hassle and expense of buying servers and equipping data centers, and paying for staff to maintain them, companies can offload their workloads to the cloud, where economies of scale around the infrastructure mean that costs are much lower.
In theory, cloud users simply pay for the resources they use, as and when they need them, without the burden of paying for hardware, or data center space. That means pricing should be straightforward, right?
Not quite: there isn’t just a single model of cloud pricing.
On-demand allows you to purchase services as and when you need them, while reserved instances work like many other types of bill, where the user forecasts what they’re probably going to need over a particular period — usually in quarterly or annual instances. The user then pays upfront, although their cloud provider may give discounts for buying services in bulk. Spot pricing is where cloud companies sell off unused processing power at a discount: companies can then bid for a certain amount of computing power at a certain price.
More of the ZDNet article from Danny Palmer