Understanding Cloud TCO

Understanding Cloud TCO 150 150 CloudGovCo

Recently, and possibly due to the pandemic, there has been a flurry of organizations looking for help in developing total cost of ownership (TCO) analyses for cloud migrations.

This is the wrong question, and the clue is in the name: Ownership. You don’t own any assets in the cloud, you subscribe, lease or rent them.

TCO comes from an on-premises mindset where you use a capital expense (CapX) to purchase an asset and defray it over a 10-year period, often with a midlife refresh. Cloud uses an operational expense (OpX) to pay day-to-day out-of-pocket costs for cloud services.

Cloud expenses should not be forecast beyond three years (five at most) because the technology, features, services and costs change rapidly. Instead of TCO use a three-to-five year Return on Investment (ROI).

Beside changing the operational model of the IT department, migrating to cloud causes financial headaches. Changing the financial model from CapX to OpX is challenging. Many organizations don’t want to show an increase in day-to-day operational expense because this is easily scrutinized by stakeholders. Capital investments, depreciation schedules and return on investment are harder to understand and easier to hide.

Finally, nobody should expect to save money in the cloud. Yes, this was the idea six years ago but in most cases cloud will be more expensive.

The business case for cloud is really about speedy development (time to market), scalability, resilience and availability — not cost. Except cost savings from reducing IT personnel which many organizations with unions are often reluctant to do. And, of course, reducing time to market is itself a cost savings and an open door to more innovation and opportunity.

To understand a few aspects of the cost issue, consider a solution on premises that has three tiers with separate servers. In the cloud it might have six or more using horizontal scaling groups plus load balancers and other services, plus data-transfer and monitoring costs, and more.

While cloud costs less than legacy solutions, hypervisors like VMware’s vSphere are very cost competitive

Now consider the case where the three tiers are virtual machines (VMs) on premises using VMware on a honking big server with maybe 32 CPUs. To begin, VMware on premises is very cost competitive with native cloud so it is difficult to show a business case on this basis.

Furthermore, this big server will be hosting many other VMs. Moving the workload for our three-tier solution to the cloud will not reduce the cost of the VMware licence or the cost of the server hardware. We will not be able to reduce expenses and take a tax break for decommissioning the hardware.

Another cost element is that by default cloud solutions should be architected for high availability using multi-zone multi-region designs. This in itself is a higher cost factor; whereas, again, this is not usually the case on premises.

Also, many organizations do not have in place effective cloud vendor-management; and do not have the capability to monitor and optimize cloud costs on a weekly basis. Nor do they have the mindset to automate to the degree desirable in cloud operations using Site Reliability Engineering (SRE).

Finally, rather than looking for cost savings, organizations should have a separate investment budget to fund a strategic transition to cloud operations over a five- to 10-year period.