Pricing Models

Cloud Server Pricing Models

It's not easy to understand the offers and the differences in price of cloud server and vps found on the Internet.
Let's try to distinguish among the various kinds of offers, or, better, the different pricing models of cloud providers.

We'll list the most used pricing models used by the providers for their cloud servers' offers.

On-Demand / Pay per Use

The On Demand / Pay per Use is the pricing model most used and it doesn't oblige the client to commit in the long run, as the price is calculated on a (generally) hourly and resource-consumption basis.

It is therefore possible to buy, for instance, a server to make tests for just 2 hours. In this case, the price is given by the product, the sum of the hourly resource cost (like CPU, RAM and disk) and by the number of hours of utilization.

Providers with the On Demand / Pay per Use model allow, in some cases, to power off the server without destroying or deleting it, thus paying a lower hourly fee, because in this case just disk space is used, no RAM or CPU resources are used.

Whom is addressed to, when use it:
Test environments and Development for a short period of time. Advised against Production environments.

Recurring fixed price

The On Demand / Pay per Use is the pricing model most used and it doesn't oblige the client to commit in the long run, as the price is calculated on a (generally) hourly and resource-consumption basis.

The recurring fixed price is another widespread pricing model. In this case the client purchases a plan that consists in the allocation of an instance (such as a Cloud Server) that has constant resources through time and a fixed price. Unlike the On Demand / Pay per Use model that have a monthly fee, or even every minute in certain cases, generally recurring fixed price plans have a duration of at least a month.

Whom is addressed to:
Testing environments and Development for large periods of time. Web Servers for projects that have a little temporal duration to handle, for instance, Events.

Reserved Instances recurring fixed price
The On Demand / Pay per Use is the pricing model most used and it doesn't oblige the client to commit in the long run, as the price is calculated on a (generally) hourly and resource-consumption basis.

To the recurring fixed price pricing family, another model belongs to, the reserved Instances model: this model is dedicated to the Enterprise world, which demands guaranteed and constant through time resource for long contractual periods.

In this case the client commits to pay for the service for the whole duration of the contract at a fixed monthly fee, while the provider commits to create a cloud server with guaranteed resources (CPU, RAM, disk I/O) that offer the certainty and tranquillity of having the same performances for the whole duration of the contract.

This kind of offer is the one the offer the client the best guarantees of proper functioning of the cloud server.

Whom is addressed to:
Production environments, Management, infrastructural Servers, Web Servers, Mail Servers, In general, server in stable Business situation.

NOTE: Difference between Allocating and Reserving.
The two pricing models, Recurrent Fixed Price and Instance Reserved Recurring Fixed Price, may seem identical at first sight. But if we look at the terminology used, we can note that in the first case we deal with an allocation of the resources, while in the second case we deal with reserved resources.
Allocating means to divide "assets" among various parts.

Reserving means to address an "asset" exclusively to a specific part.
To speak clearer, it's like booking a flight reservaton. If the we book a seat in First Class or Business Class, it is reserved: there is no chance of overbooking or other unpleasant situations.

In all the other cases, instead, the reservation (Allocation) is took into account but there is no guarantee of a sitting seat.
In the cloud world, the difference between Allocating and Reserving has a greater and more relevant impact than in the example of the booking of a flight ticket!
There are other pricing typologies less widespread like:

Variable Price on a resource consumption/On Demand basis
The price for this plan is based on the amount of resources used in a certain time frame.
It's not a widespread plan and, generally, it can be coupled with other plans (Fixed price or Variable price in time) to handle the workload peaks (Auto scaling).
The plan based on consumption resources intended as the needed resources is more common.
Variable price on a time basis
The price of this plan is calculated based on the time of utilization, instead. As a consequence, the cost of a server may decrease in the long run.

Spot Price
This model is a sort of a brokerage of the cloud providers for unused resources, resources that are reserved for a certain period. In this way, the cost is really less that than with an On Demand / Pay per Use plan.

Confused about prices?

In theory, the On Demand with an hourly fee model, the most common model, should cost more than the other pricing typologies (on equal resources and time of utilization), like the ones of the Recurrent fixed price and Reserved Instances current fixed price families.

As a matter of fact, there are several other factors that don't fully highlight this element and, actually, in some cases the On Demand pricing model is more favourable from an economic standpoint.

That happens for different reasons:

  • Cloud providers that offer Pay per Use models with hourly fee should, in theory, aim at Developers clients looking for short time Instances needed for testing and development environments. As a consequence of that, clients looking for a production environment should not aim at this pricing typology because hourly-charged cloud servers, with respect of the target they are aimed at, tend to be subject of an aggressive overbooking approach of the resources (in particular CPU and I/O disk), thus offering really aggressive prices too, with respect of the Fixed price model offers of other cloud providers.

  • The On Demand / Pay per use offers lack of services sometimes essential such as a qualified technical support via phone or those service that for a development environment may be considered as optional, but that are fundamental for a production environment. We talk about backup and system recovery services included in the fee or advanced security options.

Cloud Server as Car Sharing or long term Rent?

In general, it's like comparing "car sharing" cars that can be rent for hours and car that can be rent on a daily basis. The former are more comfortable and cheap if there is the need to have them for a few hours; the latter are the proper choice if there's the need of having a car for one or more days.
Moreover, if a car is needed for months or even years, the hourly offer will never be more convenient than a long term rent.

But in the world of the On Demand / Pay per Use another factor about the biunivocity of resources comes into play... or better, is part of a bet: a resource is available, for example, to 8 different clients at the same moment, with the "hope" that not all of them need all the power, or performances, of that very resource in the very same moment.

The Fixed price and Reserved Instances fixed price are, therefore, very similar to cars' long term rent: every client has a dedicated car and doesn't bother about someone else owning it in turn as long as the contract is valid.

Among the confusion of offers of the thousands of cloud providers on the Internet, the offer that is closer to your own business needs must be seek first, without being influenced too much by the price factor: the cheapest offers are often the ones that hide unpleasant surprises behind their appearance.

2017 Stellar is a CoreTech Srl service. All rights reserved