Calculating Total Cost of Ownership in the cloud. How to do it well and migrate with confidence

Moving to the cloud brings a host of benefits to organizations. On top of greater agility and the ability to innovate faster, the cost-optimization aspect of migration cannot be overlooked. Here as well, the cloud delivers real results. Below you will find actual data from TCO analyses conducted for various Tameshi clients from the telco and banking industries.

• Think saving up to as much as 47% on TCO thanks to the optimum use of the AWS resizable compute capacity (EC2)
• Furthermore applying tools such as the Amazon S3 Storage Lens and proper set up of managed services might bring up to 28% of TCO savings.
• A transition towards a serverless AWS Fargate, pay-as-you-go compute service results in a 49% cost reduction.

Cost of ownership calculation is a fundamental part of the assessment phase of every migration project and needs to be fully reliable to make a decision that will benefit your business. So how do we go about this process when working with our enterprise clients.

What determines your cloud TCO?

Though “know yourself” might sound like a yoga class mantra - don’t brush it off too easily. The first step to making an informed business decision is to know your starting point - i.e. the current state of your on-premise infrastructure. And “What are the costs spent on different parts of infrastructure?” is a question that companies may not ask themselves often enough. And it’s a tricky one. The lack of global procedures allowing for assigning costs to different business areas often makes matters harder. So what should an organization determine in the first place to stop guessing and start calculating its future cloud TCO? According to Tameshi’s experience, the answer to a couple simple questions is of key importance:

• What is the cost of the application?
• How much is the organization spending on its data center?

Consider the hidden costs of the on-premise infrastructure

As straightforward as these two questions might sound, In real life, answering them is actually not easy. While calculating data center costs, most organizations forget to factor in power usage, climate control, physical security, building administration etc. Hardware costs are one thing, the maintenance of the data center as a whole is another matter.

Why are these costs so important? Because they have a serious impact on TCO calculation in the cloud. Take application expenses. Assume one bare metal instance can be exchanged for a virtual instance matching the same hardware aspects. At first, this does not look like a cost-saving opportunity. That is because application costs in most cases do not include data center expenses. Once you add them to your calculation, however, the final result will likely be in the cloud’s favor.

What applications do you run and why is the cloud cheaper?

It’s fairly easy to calculate costs if the app has dedicated hardware (but even then – who is maintaining it and how much does it cost?).  What about if one server is hosting two or more applications? To accurately calculate the expenses, we should consider, among others, maintenance and labor in the right proportion.

The characteristic of the application is what determines the cost, too. An application that is not used outside of working hours in the cloud can be safely turned off generating huge savings. If utilizing an application at night is not necessary, its resources can also be limited giving the cloud a huge advantage against traditional data centers. 

What are other advantages of the cloud that cannot be implemented on-premise?

• Automated scaling - limiting costs of infrastructure during non-peak hours
• Turning systems on/off - having a system running only when it is needed
• Adding new resources on the fly
• Auto-adjusting storage
• Readiness to use multiple services

The power of combined solutions

All of the above can significantly lower TCO. That is why combining different solutions can modernize an application even more efficiently. What else is there to remember if an organization has already optimized its TCO, tailored cloud architecture, modernized applications etc.? A selection of programs and services to lower TCO even further. AWS offers many of such, most important of which are:

• Instance reservation
• Saving plans
• Spot instances (allowing organizations to use spare instances for less than the on-demand price)

If applied right, these solutions can help determine the precise cloud TCO that is significantly lower in comparison to a standard data center. Examples? With use of spot instances running cost of instances can be lowered by at least 70%, especially when used with other AWS programs (such as the Migration Accelaration Program etc.). 

Common myths around cloud TCO

Finally, to wrap it up, let’s unpack some common myths around calculating cloud TCO:

• The cloud is always more expensive as you “use somebody else’s computer” 
  No. There are only few cases when TCO in the cloud is higher than in the traditional data center. Usually, when this happens, it means the assessment phase was flawed, might not have been detailed enough or inaccurate information was provided.
• TCO cannot be precisely calculated 
  That is not true. Each cloud architecture, if based on the right, sufficiently detailed data (utilization, traffic, etc.), can be precisely calculated. Even when using auto-provisioning of infrastructure there are services and functionalities for controlling budgets and forbidding actions extending budgets.
• I know my current TCO and the cloud will not lower it
  It depends on your current infrastructure architecture, and application characteristics, to name but a few. In most cases, it is possible to lower TCO with the use of cloud services, but there are too many factors to consider to give a short answer to this one. 
• I need the same resource parameters as in a Data Center
  No, and that is because, in the cloud, there is no need to overprovision. Architectures should be designed so that autoscaling/auto-provisioning take place during peak traffic. to smoothly meet the larger demand that occurs in those rare moments.

Hopefully, this article has been helpful in understanding a little bit more about how to product your costs in the cloud.