A common misconception around cloud computing is that it’s automatically cheaper than running an on-premise infrastructure. Unfortunately, when businesses switch to (or consider switching to) the cloud, their cloud savings may not be immediately evident.
In fact, some businesses accrue more monthly costs with the on-demand cloud model than they would with on-premise systems. This begs the question: Is cloud computing really cost-effective?
The answer depends on the perspective from which you are estimating your cloud savings.
Cloud savings can be considered from two perspectives:
In this article, we’ll look at both perspectives and how they are important for understanding total cloud savings.
For businesses considering moving to the cloud, comparing the cost of the cloud to the cost of on-premise infrastructure is usually the easiest way to comprehend cloud savings. Even then, understanding your total cloud savings requires looking beyond the head-to-head cost of on-prem infrastructure versus the cloud, and considering the other intangible but significant benefits.
Below are specific areas where the cloud could save your business more money than an on-premise model.
If you are running your own data centers, there's the cost of the physical space, personnel, heating, cooling, and electricity required to operate them. There’s also a significant cost associated with the infrastructure and operational data.
By moving to the cloud, you offload the cost of infrastructure, running, and maintenance of the data center to the cloud provider.
This is one of the biggest areas of potential cloud savings. The ability to respond to spikes in demand is critical to any business and has a direct cost indication. In an on-premise model, the cost of anticipating demand can be extremely high. You would need to pay for and maintain standby servers for peak periods.
Take a nonprofit organization that supports communities impacted by natural disasters, for example.
During a disaster, more people tend to visit the organization’s site so they can donate. Of course, that's the time the nonprofit can least afford to have their website go down. But one never knows when a disaster will strike. If your organization faces unpredictable periods of high demand, how do you anticipate it?
There are two possibilities. If you're on-prem, you need to build your capacity for that maximum and always have it ready — a costly strategy. The cloud offers another alternative.
With the cloud, you can quickly scale up when you need to, and then back down when demand returns to normal. You don't have to pay to anticipate that peak load, and you can easily respond to it when necessary.
That scenario is a great example where you might pay a little bit more in raw computing power at that peak than you would if you built that infrastructure yourself. But for the rest of the time, when the infrastructure is sitting idle, that's wasted money you don't have to pay if you are using the elasticity of the cloud.
If you run your own on-prem IT infrastructure, you're responsible for securing everything from the physical space to the operating system, hardware, and software — an extremely expensive proposition. In addition, good security experts are hard to find.
In the cloud, you offload much of that responsibility. Most cloud service providers offer a shared risk model where they are responsible for securing the infrastructure and you are responsible for specific security details depending on the services you use. Overall, you get better security because cloud platforms have better dedicated cloud experts than you could potentially find.
There are also intangible savings associated with operating in the cloud. Unless your business is an infrastructure company, do you really want to be in the business of recruiting, hiring, training, and managing IT people when you could instead focus on the core activities of your business?
Otherwise expensive services, such as text-to-speech transcription, are also available to businesses in the cloud at the click of a button. This availability allows engineers to innovate and build products faster. In an on-prem environment, engineers may be more reluctant to work on new ideas because of the longer provisioning processes involved.
Data centers have huge negative impacts on the environment. They need to be maintained at a steady temperature and humidity, and they use an enormous amount of electricity. Instead of every single company trying to run these highly inefficient data centers, if you get three or four companies who can do this at hyper-scale, it's a much lower cost to the environment.
Microsoft has an impressive sustainability plan around its cloud services, aiming for a 100% renewable energy supply by 2025. For individual businesses, the likelihood of building a data center that’s 100% renewable is slim to none. As citizens of the planet, environmental sustainability is definitely an important consideration.
For businesses already in the cloud, the conversation shifts from how much they save compared to an on-premises model to cloud cost optimization.
The most important step for understanding and increasing cloud computing savings in this context is to reframe the conversation away from the absolute dollar amount per month and instead focus on the cost in terms of meaningful KPIs. This is because if your business is successful, your absolute cloud costs will increase due to the on-demand nature of cloud consumption.
For example, you'll pay more when you have a surge in website visitors than if you had little website activity. So, if you only consider the absolute cost of your cloud services, you would be missing the full picture. Instead, a better approach is to connect your cloud costs to the business value you are delivering.
Take the cost per customer of a B2B SaaS company, for example.
Suppose the company has an Amazon cloud bill of $50,000 and 1,000 customers in a particular month. That means the cost per customer for that month is $50. In the next month, if their absolute Amazon bill goes up by 10% ($55,000) but the number of customers goes up by 50% (1500), the business is doing better because the cost per customer is lower ($37). The business saved $13 per customer in cloud spend compared to the previous month.
This context is the most important thing when trying to understand your cloud costs and ways to reduce them. By having a dollar amount that's directly related to a business outcome, even if your cloud usage is going up, you can look at month over month how your business decisions affect that cost and find ways to reduce it, ultimately increasing your profit margins.
For example, a capability might be expensive to provide but isn't used by many customers. You might want to reduce your investment in that functionality. On the flip side, you might discover that a popular feature among customers has a high cost. Armed with this information, you can explore ways to re-architect that feature to deliver the same functionality at a lower cost.
In practice, this could mean switching to a different cloud service or a different usage model. It could also mean revisiting your pricing by making the expensive popular feature available in the paid tier and not in the free tier.
By connecting cloud costs to key business metrics, engineering leaders can have more strategic conversations about how to reduce costs and increase their cloud savings.
Cloud providers send bills that are prepared in their context. For example, your AWS bill shows you the number of instances used and how much they cost. But the absolute bill does not give you much insight into the performance of your business. This is where CloudZero comes in.
CloudZero takes the cost data that AWS provides, combines it with information about your application and software, and puts all that information into a context that is relevant to your business. For instance: How much do you spend on a particular feature or product, or customer segment?
Here’s a practical example: As an engineer pushes changes to production, a change might cause a performance or quality problem. With DevOps, the focus is on discovering the issue as quickly as possible (MTTK) and deploying a fix in production right away (MTTR).
It’s the same with cost. A production change can cause an unexpected cost spike. CloudZero uses machine learning to quickly identify that cost anomaly so you can manage it just like you would a performance or quality problem; get the right information to the responsible engineer as quickly as possible (MTTK), and help them understand what caused the cost and how to manage it (MTTR). This helps you to immediately analyze that change, helping to control — and ultimately reduce — your cloud costs.
Request a demo to learn more about the cloud cost intelligence that CloudZero provides.