Total cost of ownership, aka TCO, has become the data point we want to understand most as cloud computing moves into our enterprises. Better put: We endeavor to pinpoint the costs of using public and private clouds in the enterprise over the next three to five years and the value (if any) it will return.
This may seem like an easy number to calculate, but in reality the cost models are complex and ever changing.
Here’s why:
- Most enterprises use more than one public cloud (multicloud), and the workloads running on each cloud change randomly. TCO planning requires working with specific workload profiles and making assumptions about which workloads will run where, but both factors will be more dynamic than originally thought.
- Prices shift constantly, typically downward, and assumptions about costs can be off in six months or less.
- Expenses for talent are not often considered -- but they're significant. Most organizations have to makes changes around the use of cloud computing, including hiring developers, architects, admins, and so on, which increases TCO greatly.
- Not everyone dials in security and governance until after implementation. This cost normally accounts for 20 to 30 percent of TCO.
The end result is sticker shock when understanding the real “all in” costs, including expenses we didn't originally consider, as well as dealing with dynamic workloads and changing prices. Most enterprises are way off the mark, and they seem to be getting worse at understanding TCO.
Enterprises should build cloud computing costs into their spending models, the way they would for telecom or other utilities, with a variable portion and a fixed portion. Enterprises don’t have much experience with cloud services and thus don’t consider organizational costs, security costs, networking costs, and so on, and they blow up the budget.
The only way to fix this is to first learn what’s changing and what’s not. Understand the impact of the use of cloud computing, including the costs and benefits. The costs are justified if the benefits are there.
Make sure to monitor the consumption patterns over time and incorporate the usages changes in the cost models, then adjust accordingly. For now, that’s the best way forward.