Muhammed Abiola
/usr/share/blog

Follow

/usr/share/blog

Follow
The Economics of AWS Cloud Computing

Photo by Taylor Vick on Unsplash

The Economics of AWS Cloud Computing

Muhammed Abiola's photo
Muhammed Abiola
ยทMar 17, 2023ยท

9 min read

Introduction

Cloud computing is the on-demand delivery of compute, databases, storage, networking, and other resources, over the internet via pay-as-you-go pricing. And whether you're using it to run applications that share vanishing photos with millions of mobile users, or to support critical healthcare operations, cloud service providers such as AWS offer rapid access to flexible and low-cost IT resources.

These facilities matter, because, with cloud computing, you never need to make large upfront investments in hardware or spend a ton of time on heavy lifting or procurement. Instead, you can provision just the right sort of resources you need to power your ideas and scale according to demand, paying only for what you use. This is why it's such a step change.

AWS was founded in 2006, initially only offering Cloud Computing services via EC2 (we'll be talking about this soon!) - and has since diversified more than 200 offerings across multiple verticals such as healthcare, machine learning, IoT, blockchains, and more. In doing so, becoming a critical aspect of modern business operations today, offering a wide range of benefits such as scalability, economies of scale, and increased efficiency.

This article aims to provide a brief but cogent overview of the economics of AWS Cloud computing, exploring the benefits of moving to the cloud and how it can reduce costs for businesses of all sizes.

Benefits of moving to AWS

According to several studies, the most efficient and advanced data centers around the world are operating at less than their full capacity, with no more than 60% of their potential at any given time being utilized, leaving 40% of compute resources to gather dust, with no ability to harness their potential. This is largely due to how applications are architected. Activities performed in the cloud (known as workloads) are highly variable and cyclical, and in an on-premises environment, it's important to provision for peak demand. The rest of the time, these resources run idle. This comes after a timeous procurement process and high initial capital expense that can't be traded. This wasting of capacity created an opportunity that led to greater efficiencies, cost savings and performance in data center operations, leading to one of the greatest business outcomes we've ever seen, growing to a valuation of about a trillion dollars in under 20 years.

Using cloud infrastructure, we're able to drastically reduce procurement time, from weeks to minutes, without an upfront expense, allowing enterprises, startups and small businesses access to the building blocks they need to respond to quickly changing business needs. Businesses are now able to change their compute instances as needed, scaling them up or down as needed, or even being able to sell them, so they can reinvest in their core offerings.

Financial Metrics

Operating Expenses (OpEx) and Capital Expenses (CapEx) are some of the most important financial metrics to consider when comparing on-premises infrastructure or a virtual data center, to the cloud, and they're used in measuring the total cost of ownership. Let's take each of these terms sequentially.

Operating expenses refer to the ongoing expenses associated with running a business, such as personnel, power and maintenance costs. For an on-premises configuration, OpEx would include costs such as personnel salaries for server administrators, support teams and managed services, as well as the costs for training and education. It's worth noting that moving to the cloud would lead to significant savings in labor costs, through the reduction of in-house personnel needed to manage and maintain hardware and infrastructure.

Capital expenses, on the other hand, refer to the one-time expense associated with acquiring property, plant and equipment such as storage facilities, bandwidth and connectivity, switches, and load balancers. CapEx would include the upfront costs of procuring and installing a data center, as well as the costs of hardware and the infrastructure needed to run and maintain it. Here, there's also the treatment of taxation, which varies widely across jurisdictions and may impact the total cost of ownership of on-premises infrastructure. CapEx is a deductible expense for tax purposes. However, this treatment varies depending on (1) the sort of asset, (2) the jurisdictional tax code and (3) the industry the business is in. Cloud computing services given their pay-as-you-go nature, eliminates those upfront costs and greatly simplify tax considerations.

Now, the total cost of ownership (TCO) is a financial metric that takes into account both the OpEx and CapEx to provide a comprehensive view of the associated costs of acquisition, usage and disposal of products and services over their lifespan, allowing customers to model the cost of running workloads in the cloud, and compare them to an on-premises configuration.

Creating a data center cost model can be challenging for various reasons. One of which is the difficulty in accurately estimating the cost of power and cooling required for each server and service, as it may vary significantly depending on a few factors including geographic location, the sort of cooling system used, and power requirements. We also need to think about how complex the IT infrastructure is (servers, storage, networking), how to expense rapid changes in technology, and the variability of our workloads. These factors altogether make it difficult to provide an accurate estimate of TCO, which is crucial for businesses to make informed decisions about their infrastructure investments.

While there are tools and methodologies available to help with this process, such as data center infrastructure management (DCIM) software that measures, monitors and manages utilization and energy consumption, and industry benchmarks for data center efficiency and cost, AWS ultimately has a lower total cost structure than on-premises infrastructure, whether it be virtualized or physical.

Some benefits are:

  1. It runs at a much higher scale: AWS has availability zones (data centers clustered together) all around the world, and each data center has up to 15,000 servers - offering potentially millions of instances. This leads to economies of scale and reduced per-unit costs.

  2. Utilization in the cloud is much higher: Services like AWS have millions of services running at any one time. Some with high utilization, and others with low utilization. AWS aggregates non-correlated workloads, allowing customers can buy instances on the spot market - which is, essentially, all the extra capacity existing in their data centers - at a tenth of the cost, driving up total utilization.

  3. Amazon-specific hardware designs: AWS operates at an unprecedented scale, and some of the hardware provisioned by AWS isn't available anywhere else on the market. Customers have direct purchasing access to CPU, memory and disk options, where AWS controls the hypervisor and net protocol layers, enhancing overall performance and security.

Let's consider a simple example to calculate the total cost of ownership for an e-commerce site, that uses a data center to host its website against the migration to an AWS cloud configuration.

Cost CategoryOn-premisesAWS Cloud
Hardware$5,000$0
Software$365$0
Cloud Infrastructure$0$3,000
Website Development$2,500$2,500
Maintenance$800$800

Under an on-premises model, the one-off hardware cost for our compute, storage and networking is $5,000, there is also a yearly software-based cost of $365 in an on-premises model. Using AWS however, there are no upfront hardware costs - since compute is provisioned using a pay-as-you-go model. Similarly, since we can operationalize our servers, we can manage email, storage, and compute in a pay-per-use way.

Further down, we see that an AWS cloud configuration incurs a sizeable cloud infrastructure cost of $3,000 a year. Both the on-premises model and AWS cloud configuration have website development costs of $2,500 and maintenance costs of $800.

The table provides a simple example to calculate the total cost of ownership for an e-commerce site and illustrates that migration to an AWS cloud configuration can result in significant savings on hardware and software costs but incurs additional costs for cloud infrastructure. We observe that AWS leads to significant savings in up-front costs, but those begin to creep up over time as we use various AWS-based products and services.

We can simplify this by separating these costs into CapEx and OpEx:

Cost CategoryOn-premisesAWS Cloud
Hardware$5,000$0
Website Development$2,500$2,500
Total CapEx$7,500$2,500

CapEx for the on-premises configuration is $7,500 which includes the one-time purchase of hardware for $5,000 and website development costs of $2,500. CapEx for AWS however is only $2,500 for our website development costs. The AWS cloud configuration has a lower CapEx compared to the on-premises model by 67%.

Now, comparing our operating expenses for each configuration:

Cost CategoryOn-premisesAWS Cloud
Software$365$0
Cloud Infrastructure$0$3,000
Maintenance$800$800
Total OpEx$1,165$3,800

We see that the on-premises model has a total OpEx of $1,165, consisting of a software cost of $365 and a maintenance cost of $800. Meanwhile, the AWS cloud configuration has an OpEx of $3,800, which includes a cloud infrastructure cost of $3,000 and a maintenance cost of $800.

While the on-premises model incurs a higher CapEx, it has a significantly lower OpEx compared to the AWS cloud configuration. In contrast, the AWS cloud configuration has a lower CapEx, but higher OpEx, primarily due to the cloud infrastructure costs. A difference of about 226%!

Given that we've plugged in all our variables, we can calculate our TCO for each case:

On-premises:

Cost CategoryCalculationCost (per year)
Capital ExpenditureHardware + Website Development$7,500
Operational ExpenditureSoftware + Maintenance$1,165
Total Cost of OwnershipCapital Expenditure + Operational Expenditure$8,665

AWS Cloud:

Cost CategoryCalculationCost (per year)
Capital ExpenditureWebsite Development$2,500
Operational ExpenditureCloud Infrastructure + Software + Maintenance$3,800
Total Cost of OwnershipCapital Expenditure + Operational Expenditure$7,300

We can see that despite the massive increase in OpEx, an AWS cloud solution offers significant cost savings compared to the on-premises solution, with a 27% savings in TCO. Note that this is an extremely stripped-down and simplified model. Ideally, you want to consider multi-year trends, and the cost of each service for the workloads you're running. Also, operating expenses are ongoing and variable. They're incurred in the day-to-day operations of a business, such as utilities, salaries, and administrative costs. Over time, as you scale, these are likely to increase, while CapEx will remain relatively stable.

It's a good idea to reduce CapEx, mostly because it frees up capital that can be used for other research and development investments or growth opportunities. While OpEx costs may increase over time, the goal is to keep them as low as possible through various cost optimization strategies, to maximize profitability.

Conclusion

Cloud computing has become a critical aspect of modern business operations today. Here, we provided a simple overview of the economics of AWS cloud computing, exploring the benefits of moving to the cloud and how it can reduce costs for businesses of all sizes. We discussed the financial metrics of Operating Expenses (OpEx) and Capital Expenses (CapEx) and how they're used in measuring Total Cost of Ownership (TCO).

Ultimately, moving to the cloud offers a wide range of benefits such as scalability, economies of scale, and increased efficiency, making it a smart investment for businesses looking to optimize their operations and reduce costs. You can explore tools like the AWS Migration Hub and Amazon's Migration Evaluator to provide insights into a migration process.

Thanks for reading!

ย 
Share this