I’ve been reading through a book about FinOps. It appears to be a topic which is getting increasing attention. I expect that most of it can be boiled down to some key principles.
- No longer about costs being in the hands of procurement / purchase orders
- Bringing the idea of cloud costs to every part of business and engineering
- Being able to take ownership of usage (not about rates)
- Runaway cloud costs can threaten a business’s survival!
- The business needs a plan
- Cloud is rarely about saving money, it’s about increasing the potential for innovation and competitive advantage
- It’s about making money rather than saving money
- Requires real time data to make data-driven decisions
- Outcome: efficient cloud costs balanced against the speed, performance and quality, availablility of the services
- Outcome: better decisions while moving quickly: frictionless conversations
- Three parts: real-time reporting + just-in-time processes + teams working together = FinOps
- Prius effect: immediate feedback loop
- E.g. a story from a friend: the cloud had databases growing exponentially which added an extra cost of £50,000 per month! which was unbudgetted.
- In another example: $200k on services that weren’t being used (and it took 3 hours to turn off)
- In another example, the developers had zero idea that the cloud services they were using even had costs! And had no experience of managing costs.
- FinOps should start as soon as migrating to cloud … not when panicking at the sight of eye-watering fees (avoid the problem)
- Requires as ever … a cultural change
Principles:
- Teams need to collaborate
- Decisions are driven by the business value of cloud
- Everyone takes ‘ownership’ of their cloud usage
- FinOps reports are accessible and timely
- A centralised team drives FinOps
- Take advantage of the variable cost model of cloud
Next steps:
- First step is visibility of cloud spend to relevant teams in a timely manner so that quick decisions can be made
- Cultural change / training in cloud / engage finance
- Grow incrementally
- Mature teams anticipate cost overruns are prevented via thoughtful architecture design
Key metric: unit economic metrics to enable data driven decision making by engineering teams
Definition
Cloud:
- OSSM
- On-demand
- Sacalable
- Self-service
- Measurable
Problem:
- Suddenly engineers can spend company money with the click of a button
- Up to 1.8tn will be spent on cloud per year
- There are natural constraints to costs on-prem, but not on cloud!
- Need to create that constraint
A FinOps team
- Cross functional covering needs of
- Executive buy in
- Engineering and development
- Finance
- Procurement and sourcing
- Product and business teams
- FinOps expertise
Report to CTO, CFO, CIO
What kind od metrics does this team need:
Engineers want
- Priority
- Speed
- Performance, uptime, resilience
- Quality, features etc
Finance
- Forecast and predict spending
- Allocate spending
This is as far as I have got. I think visibility and dashboards is probably key!