My postings reflect my own views and do not necessarily represent the views of my employer, Accenture.
Is Cloud cheaper than on-premise?
Half the room might say yes, and the other half, no. It obviously depends, but on what? Without offering a perfect model, the problem statement includes: 1) designing cost efficient Cloud architectures, and picking the right Cloud services, 2) picking which features and tiers to enable within those services, 3) rightsizing the volume and capacity if this applies, and 4) buying what you need effectively in terms of commitment plans and reservations offered by the CSPs. Getting all of that right, on a continual basis is FinOps. It is the negative entropy preventing ballooning Cloud costs. Like everything else in the realm of Cloud, it’s evolving very quickly. FinOps is engaged in an arms-race. It is a fair guess that FinOps will look very different in five years time.
Focusing on ‘Who’
The two safest bets about FinOps evolution would seem to be:
- Increased automation and better tooling, reducing the FinOps workload
- Stabilisation of a default operating model and methodology, making FinOps more widespread
But who is expected to drive this evolution? The application team? The central IT function? Cloud Service Providers? Paid third parties? Open-source communities? Everyone? Who is most on the hook, and who is just along for the ride?
First I take a look at the key players… how they might shape the evolution of FinOps (+ve points)… what their limitations might be (-ve points)… and their overall personalities in the story. Second, I make some predictions i.e. the title of ‘FinOps in 2025’, but the date isn’t literal!
The Players, Their Personalities
Application Team — “The Busy Parents”
All roads lead to Rome.
- Primary decision makers for infrastructure and application architectures. This will determine the major architectures that FinOps needs to have an answer for
- Best able to prioritise/ authorise FinOps opportunities based on context. This will determine how much of the implementation is automated, and how much needs human approval
- Expect patchy adoption if the problem is left only with the Applications teams, since standardisation requires platform-level thinking
- Empowerment is becoming overwhelming, and not everything can be the #1 priority i.e. Ops > DevOps > DevSecOps > DevSecFinOps > something gets ignored
Central IT Function — “The Brokers”
In bigger companies, probably also supported by advisory partners. Note: per FinOps.org principles: “a centralised team drives FinOps”.
- Drives standardisation and common set of best practices from the platform level, downwards
- Best placed to play the broker role between all external players and the Application Teams
- Drives sharing and economies of scale i.e. cross-team capacity commitments and shared utilisation
- Needs empowerment and buy-in from application teams or else can become paper-based analysis
- Difficult to drive innovation in tooling or methodology given capital and skills requirements; prefer to fast-follow
Cloud Service Providers (CSPs) — “The CEOs”
E.g. Google, Amazon, Microsoft, etc.
- Sustainability as competitive edge i.e. FinOps > less consumption > smaller carbon footprint > and lower costs > passed back to customer
- FinOps support as competitive edge i.e. attract and maintain customers based on FinOps cooperation
- Pattern of offering managed open-source platforms, should a major FinOps open-source tool emerge
- Pattern of baking in automation to CSP specific products, which may include FinOps levers, to increase vendor lock in
- Revenue reduction if always giving the best and cheapest deal
- Can’t easily offer hybrid or multi-cloud perspective or else lose further revenues (CSPs will offer a multi-venue control plane e.g. Anthos, but will organisations let them self-optimise?)
- Unlikely to drive methodology innovation as this is outside the CSP side of shared responsibility model
Paid Third Parties — “The Entrepreneurs”
E.g. Apptio, Zesty, etc.
- Impartial hybrid and multi-cloud solutions based on AI/ML that are beyond what can be developed in-house
- Outsourcing of problems, i.e. can simplify customer processes and skill recruitment needs, and offer outcome based models
- Everyone needs paying, but maybe only by a little if there is sufficient scale
- Most likely to be shut out by customers not engaging their optional services, or by CSPs via acquisition or imitation if they feel threatened
- Unlikely to be opinionated in terms of methodology as this may reduce potential customers and revenue
Open-Source Communities — “The Democrats”
E.g. Linux Foundation, Apache Foundation
- Open-source is where it’s at right now, and there seems to be an eventuality that the community will take the best of what everyone else produces, and democratise a version. Very warming
- Most likely to house and curate innovations in the methodology space, as per the trend with Agile, DevOps, GitOps, etc. Paid methodologies are almost unthinkable
- There doesn’t seem to be a front-runner FinOps tool/ platform with a committed community, yet. There are plenty of small technology-associated projects appearing, but nothing trying to solve “The AWS problem” (or Google, Azure, etc.). There is much focus on platform-wide observability which may lead to adapted use cases
Predictions For 2025
- This isn’t really a problem for the Application Teams to solve. Sure, it’s much much better if the Application teams are economically-aware and make sensible choices the first time round, but the FinOps discipline needs to solve for the worst case and protect the organisation from unaware or uncooperative Application Teams. This doesn’t mean that the Application Teams are any less important in the evolutionary process or debate, quite the opposite, but they should expect to get plenty of help and probably not lead the way
- The major relationship to get right now/ first is between the Application Teams (“Busy Parents”) and Central IT Function (“The Brokers”), to establish initial best practices and the FinOps-culture. This is at the core of all emerging operating models. The strength of this relationship may determine the extent to which 1) Application Team (especially developers) are protected from the noise of fast paced evolution, and 2) how easily organisations can adopt new methodologies and technologies as they arise and mature
- CSPs have a bigger role to play in the future, and they are experiencing the biggest and most confusing set of competitive forces. From a technology perspective only the CSPs offer a ‘fix at source’ solution. But for each % efficiency offered back to customers, billions in lost revenue needs to be offset by increased customer retention and customer-conversion. Technologically, expect to see continued evolution in serverless, auto-scaling, multi-service right-sizing, and AI driven optimisation. But the more significant changes may be economic, geared to simplify or even automate processes around reservations, commitments and pricing plans
- The assumed/ expected rise in hybrid-cloud and multi-cloud architectures is an opening for Paid Third Parties (“The Entrepreneurs”), and Open Source Communities (“The Democrats”) and we should expect a glut of start-ups/ new projects and acquisitions. But organisations should pick carefully. To name only a few in the paid space: Apptio/ Cloudability, SoftwareOne, KubeCost, Zesty, Densify, ParkMyCloud, Blissfully, etc. — my point — there are lots of them, and likely to be many more. But it’s early days. Organisations that focus on just one tool, or too many, may have rework processes in the near future. Tactical exploitation may be best for the immediate future
- Sustainability is not really part of the FinOps movement right now, but could become a CSP owned problem. The focus is on cost management from the customers’ perspective, which makes sense. Whilst there is growing pressure on CSPs to be greener and more sustainable, the general mood has not connected the dots to argue: ‘FinOps > less consumption > smaller carbon footprint > and lower costs > passed back to customers’. If this argument prevails it may well be against the will of the CSPs
Please reach out to discuss any of the ideas, I’d love to help.
