Cloudthread Y Combinator
April 18, 2023

Cloud costs: opposite CTO perspectives

CTO perspectives on cloud cost waste reduction: abstract costs away from engineering/product vs culture of cloud cost accountability

At Cloudthread we talk to a lot of CTOs about cloud cost waste reduction. I’ve noticed two fundamentally different perspectives.

These two perspectives represent extremes, most organizations lie somewhere in between, but all organizations need to consider this tradeoff.

Perspective 1: “I don’t want engineering/product to worry about cloud costs at all. We’ll optimize with committed use discounts.”

CTOs in this camp don’t want to burden engineers with cost consideration given the opportunity cost of their attention/time. Engineering time is one of the most valuable things at a software company and they already need to juggle feature development, stability, security, etc.

This perspective goes hand in hand with a growth over profitability mindset. The organization gets more value from engineering time dedicated to accelerating the product roadmap than it does from optimization. Many organizations that were singing this tune in 2021 are shifting their views given the current macroeconomic climate.

In Cloud FinOps terminology, this perspective is summarized as: CTOs want to do rate optimization but not consider usage optimization at all.

Pros of Perspective 1

  • Engineers get zero “distraction” from cloud costs
  • Don’t need to perform granular chargeback/showback in order to attribute costs to owners

Cons of Perspective 1

  • You’re inevitably leaving money on the table in the form of cloud cost waste
  • Any central FinOps team is paralyzed to act on usage optimization cost efficiencies without getting Engineering input and sign-off (can that EC2 instance be rightsized? Can this elastic IP be released? Can that idle RDS instance be shut down?)
  • You’re getting committed use discounts (whether enterprise discount programs, RIs, SPs, CUDs, etc) on inefficiently used resources
  • It’s hard to walk back from this culture to one of engineering cost ownership

Perspective 2: “I want cloud cost ownership to be a distributed responsibility for each engineering/product owner”

CTOs in this camp believe thei engineering/product teams care about the bottom line and, if equipped with the right data, they can have ownership of costs and drive infrastructure cost efficiency”. These CTOs believe in the prius effect: exposed to cloud cost data and ways to reduce cloud costs, engineering teams will take remedial action.

This perspective goes hand in hand with a “you build it, you run it” culture that includes cloud cost ownership. This organization has invested in cloud cost attribution (with cost allocation tags) and wants cost waste to be reviewed on a regulard cadence and egregious cost anomalies to be surfaced and acted on by owners.

Pros of Perspective 2

  • Cloud waste is actively managed
  • Leadership is (ideally) confident in cloud cost efficiency
  • Committed use discounts are applied to properly sized resources
  • Central FinOps team gets engagement when surfacing cost efficiencies to engineering/product owners

Cons of Perspective 2

  • There is some % of engineering bandwidth dedicated to cloud cost reduction
  • Need to perform granular chargeback/showback in order to attribute costs to owners
  • Need a platform to surface and attribute cost saving opportunities

Incentives and Tradeoff

Stability is something that is an inherent consideration for engineering/product because it affects the user experience. Cloud cost inefficiencies aren’t reflected in the user experience. Cloud waste reduction comes from adequate visibility, processes, and incentives to facilitate a culture of cost ownership.

“Show me the incentive and I will show you the outcome” - Charlie Munger

Creating a culture of cloud cost ownership takes an upfront investment to provide the required visibility and ongoing attention for engineering/product owners to maintain cost efficiency.

It takes a leadership level decision on culture and incentives around cloud costs in order to influence decision making and prioritization.