Determining the potential costs of cloud computing can be as complex as the technology itself. Not only are you faced with the different pricing structures of cloud providers, you must find a reasonable way to estimate the resources that you will need in the future. And despite your best efforts, there is no guarantee that your calculations will be correct. That’s because of the variable cost architecture of the cloud.
Of course, you want it to be more than a guessing game.
One way to estimate future usage is to look at what you’ve used in the past. For example, if you can look at your electric bills for the past year, you can get a pretty good idea what your expenses will be over the next few months. That presupposes that you are not making any drastic changes in your electricity usage.
Herein lies the problem. Suppose that you have a website that markets a certain product. Because you have hit the market at the right time with excellent promotion, the website takes off. Thanks to the scalability of the cloud, sufficient resources are automatically ramped up to meet the increased demand. This is good, because the more popular your site becomes, the more sales are completed. On the other hand, your cloud computing costs may go through the roof.
Anticipating the need for resources is the key to calculating cloud computing costs. Those resources include compute, storage, and network. You may find that your cloud computing budget needs to be as flexible as the cloud resources that you are using.
From CAPEX to OPEX
There are plenty of models and methods out there for budgeting projects and managing cost centers. Unless you manage a cost center yourself, you may never have had to deal with the financials related to information technology. IT professionals have enough to do to keep up with the latest practices in their particular areas of expertise. But as careers develop, even the best of us find that we have become wrapped up in matters of budget and finance. Rather than espousing a particular financial method, let’s look at tech budgeting in broad strokes.
The traditional computing model was dependent on significant capital expenditures (CAPEX). The one-time purchase of hardware, software and licenses meant that companies had to squeeze as much work out of these resources as possible throughout the life cycle of these platforms. There was always a strong focus on the maintenance and configuration of proprietary machines. This meant that vendor support was essential to keeping systems current and healthy.
Just as a reminder, capital expenses related to the traditional data center include: servers, switches, routers, firewalls, storage, load balancers, software and licenses. The technical expertise required for installation and maintenance of these systems also costs money in terms of salary, wages and subcontracting expenses.
The advent of the cloud changes all that. The cloud allows businesses to focus on services rather than the underlying technologies. That means that IT budgeting moves away from CAPEX and pays for the cloud as a recurring operating expense (OPEX). In fact, many IT costs become difficult to track because they move to other cost centers. For instance, cloud financial services may now be covered by a company’s finance department. What were once considered IT services in the old data center model have become known by the functions that they are addressing rather than the technology that runs them.
So who pays for cloud services? It could be the sales department, or finance, or operations. Increasingly IT departments may play a lesser role.
Cost Centers in the Cloud
The three cost centers of the cloud – compute, storage, and network – provide the outline for the calculation of cloud costs. Author Joe Palian of Expedient addresses these three areas in answer to the question “How Do Cloud Providers Decide How Much to Charge?” But these broad areas don’t account for everything. The options and variables related to cloud usage can make predicting costs something of a guessing game. But the main components give us something to work with.
The costs for compute depend entirely on what you’re going to do with it. How much processing power is required for your computing projects? A common way to deal with computing capability has been to purchase more than you need. Better to have too much than too little, as they say.
But the scalability of cloud computing means that you can take a much different approach. If you have a temporary project that is entirely contingent upon the response of website visitors, you can use cloud computing to scale up automatically and scale back down just as quickly. The computing systems that you set up with cloud providers can purchase for on-demand usage or for a fixed period of time.
The parameters involved in selecting a CPU include the operating system and the expected usage (in percent). Cloud providers will then calculate the cost of CPU based on their cost per gigabyte (GB) of virtual RAM.
Advancements in storage technology have brought down prices considerably. It is no longer necessary to have dedicated hardware for each client or project. Virtual disks have replaced their physical counterparts. The same scalability in the compute sphere applies to storage as well. Storage is calculated in units of GB of virtual disk.
While these are the main cost centers, not everything that a cloud provider offers will fit neatly in these three categories. Each provider packages their offerings in different ways. It might even seem like comparing apples to oranges when putting one service up against another.
A further complication in calculating cloud computing costs is the fact that most companies are turning to multiple providers to handle their cloud computing needs. The research firm IDC found that 84 percent of IT executives expect to use multiple clouds from different providers. It’s not enough to determine what your favorite cloud company charges if you are going to spread out your cloud presence to more than one company.
Kip Compton of Cisco blogged about the complexities of the multicloud. And he suggested several questions to ask. How will you manage the different providers? How will you keep everything secure? How will you balance cloud resources with those you have on premises? But the real question that many companies will want to know is this: How much will it cost?
Companies like Cisco and Nutanix offer support for the management of a multicloud strategy. They can even use automated processes to redirect traffic to more efficient or economical resources – even if it means dynamically moving to another cloud provider.
Using the Calculators
Despite your best efforts, you may have a hard time estimating cloud costs using your spreadsheet or other bookkeeping tools. You’ll be happy to know that the cloud providers have done the work for you. Here are some calculators taken from a list compiled by TechTarget:
Amazon Web Services: AWS Total Cost of Ownership (TCO) Calculator
Google Cloud: Google Cloud Platform Pricing Calculator
IBM Bluemix: IBM Cloud Pricing Sheet
Microsoft Azure: Microsoft Azure Pricing Calculator
VMware vCloud: VMware vCloud Air Pricing Calculator
CenturyLink Cloud: CenturyLink Cloud Price Estimator
Profit Bricks: ProfitBricks Virtual Data Center Calculator
Interoute: Interoute Pricing
Rackspace: Rackspace Cloud Calculator
Cloudops and Workloads
In an article for InfoWorld, author David Linthicum suggests another approach to calculating cloud costs. He talks about the new buzzword “cloudops,” which he defines as “the ability to operate workloads, including both applications and data, once they get to the public cloud.” He proposed what he called a simple formula:
Cloudops Cost Per Year = ((NW*CW)*COM)+((NW*CW)*SR)+((NW*CW)*MR)
The variables include number of workloads (NW), complexity of workloads (CW), security requirements (SR), monitoring requirements (MR), and a cloudops multiplier (COM). Whether you consider this formula simple probably depends on your own mathematical abilities. Filling in the variables could be a challenge.
But Linthicum tries to ease the minds of his readers:
Don’t worry about this specific formula; its general nature will likely result in an incorrect calculation when you first use it. Instead, consider it a starting point, then customize the formula and your definition of the correct values over time to end up with a formula you can wield with assurance.
Everybody’s cloud computing needs will be different. Linthicum thinks you should come up with your own formula.
This is where the variable cost architecture comes in. The variables in the formulas above are left for the cloud user to determine. A cloud provider can tell you how much they might charge for a certain number of gigabytes of data. But they won’t tell you what that number should be. That’s up to you and your project designers. The provider won’t be able tell you how many servers to request, which type, or how many GB of virtual data you might consume. Likewise, they can quote you a price per GB of data transfer, but it is not their place to say how much that should be.
In determining your potential resource usage, you might want to put together your best technical minds into a room with your best financial personnel. And you may have to change your way of thinking. You are no longer purchasing data center hardware and software. You are matching your IT requirements with a flexible and scalable cloud architecture. If you only think in terms of hardware and software, you may never be able crunch the numbers properly.
There are some strategies that might help. Rather than searching for a precise number, what about creating a possible price range? What’s the highest price you could live with? What do you think would be the minimum required? Perhaps setting high, medium and low usage points would be beneficial. And of course starting small in any migration to the cloud is generally a good idea.
Moving to the Cloud
If you’ve never managed cloud resources before, it can be a bit intimidating. Techopedia addressed the transition in an article called "What Moving an Idea to the Cloud Actually Entails." There are checklists and strategies and lots to think about. The whole concept of a data center has been transformed in recent years.
When calculating costs, you will have to determine how your current IT infrastructure matches resources in a cloud environment. Will you try to move everything at once, or will you take a gradual approach? It’s better to start small.
Even so, it might be difficult to predict cloud costs based on how things worked in the data center. Remember, cloud computing has changed the orientation from system maintenance to application assurance. The apps you run on the cloud are agnostic about the infrastructure that runs beneath them.
Controlling Cloud Costs
Some of us are old enough to remember what was known as the “energy crisis” back in the 1970s. Because of a complex set of international issues, gasoline stations ran out of fuel as drivers tried to fill up their tanks. The U.S. government put regulations on thermostats for the heating and cooling of government buildings. And the speed limit on national highways was limited to 55 mph. Sometimes controlling usage is the most important thing.
That principle applies to the usage of cloud resources. Unless you have an unlimited budget (which no one does), you will need to find ways to keep costs down by capping usage where necessary. The pay-as-you-go nature of cloud computing requires sound financial management. Particularly when starting out, you will want to keep a close eye on what resources you are using.
The good thing is that it is possible to configure cloud devices to shut down automatically when not in use. Are your cloud resources not needed after hours or on the weekend? You can set up your virtual servers and network to operate at specified hours.
Techopedia has addressed the subject in detail in a white paper called Controlling Costs in the Cloud: 8 Things Every CIO Must Consider. It would be a good document for anyone interested in cloud costs to download and review.
Total Cost of Ownership
One mistake that IT managers might make when dealing with infrastructure is to ignore the concept of total cost of ownership (TCO). Calculating TCO is not just about one-time costs. You need to think about what the expenses will be over time.
Cloud providers offer discounts for long-term commitments. On-demand or as-needed services give you plenty of flexibility, but they may not give you the best cloud prices. Commitments for one, two or three years will help you lock down prices and make your costs more manageable. Such arrangements can give you discounts of 70 percent or more.
The thing about cloud computing TCO is that you no longer have to worry about the life cycles of equipment or other data center infrastructure matters. In fact, you may only need cloud services for a short-term project – in which case, you’ll be happy to use usage-based resources rather than a purchased infrastructure. One company recently had such a project. The manager reported how they were easily able to scale up to meet resource demands and come right down as the needs diminished.
Let’s face it: So much of the work done in information technology is project based. Even those things that might be considered routine maintenance procedures, when viewed through the eyes of a project manager, share many of the same characteristics of any other project. When calculating the costs associated with your cloud computing needs, you should establish a plan that encompasses all aspects of your cloud project. And that goes beyond projecting gigabytes of virtual RAM.
When defining scope, you should brainstorm anything that might be needed, such as:
What about the technical expertise to implement and manage your cloud resources? Do you have those in-house, or will you be working with contractors? How many hours of work will be required from those personnel, and at what pay rate? Will you need to pay for licenses for software that may be installed on the cloud? Are there other fees involved? What about your own salary and expenses while on the project?
In the end, your final calculation for anticipated cloud computing use might be considered nothing more than an educated guess. Along with determining costs, you will need to find ways to save money where you can. This means looking for possible discounts, turning off cloud devices when not in use, or making longer commitments when it makes sense. And your method for calculating total cloud costs may be one that you develop on your own. You know the variables better than anyone. Whether you come up with your own complex algorithm or develop robust accounting worksheets, you will have to keep a close eye on these costs. Better to control them now than to let them spiral out of control down the road.