Cost in Cloud - a FinOps Lens

Cost in Cloud is an area of many notions and few explanations.

Holistically, the question about Cloud boils down to three aspects – Speed, Quality and Cost.

The bad news is that all three parameters cannot be at best together. The good news is that in Cloud, this decision of prioritizing which parameter should be key is granular at several levels. Some examples to explain this granularity -

• A high compute instance running heavy backend operations can be refactored into server less functions. This will save costs, without compromising on quality, but may not be a speedy implementation.
• Going in for cloud native database will provide needed quality and high change velocity but will not save you as much cost.

Thinking about Cost in Cloud needs a threefold approach.

Strategy

The granularity of Cloud costs varies by service with every Cloud provider. A well weighted decision tree should be followed to arrive at the cheapest per unit prices for every project in question.

Engineering

During both Build and Run, the Cost mindset must be always on. Architecture, development and specialist activities on a given service should understand how that service is billed and build a solution that will keep the run cost low. Repeatable operations should be automated via IaC (Infrastructure as Code).

Finance

Cloud has an inherent need to keep a tighter eye on the spend. There is a need to evolve financial processes from monthly bookkeeping to near real time monitoring and alerts. Industry wide standards should be established for setting thresholds and budgeting, and roles should be in place to enable real time decisions to alter Cloud spend when needed.

Overall, every cost related activity in Cloud can be categorized as Cost Avoidance or Cost Optimization.

Cost Avoidance is using less. 

Some examples are -

• Resources are not over provisioned or under provisioned
• Automated shut down of non-production servers outside business hours
• Backing up systems only when needed
• Usage of right Cloud service
• Cheaper Cloud regions when performance and data sensitivity is not a constraint
• Right architecture to support High Availability and Disaster Recovery with native cloud features
• License cost avoidance using Cloud native services or Pay as you go models

Cost Optimization is paying less per use.

Some examples are -

• Negotiate discounts
• Setting up reserved instances or committed use discounts
• Automation via IaC (Infrastructure as Code) for repeatable scenarios

If you think Cost in Cloud is a topic you are struggling with, try to think if you have the necessary knowledge workers to help you achieve your target, whether there are right processes in place for the new way of working, and right tools are in place to monitor and manage the spend. This is because Cost in Cloud, outside of the technology, is also disruption to existing ways of working that any enterprise going towards Cloud will have to address.