Financial Liberation: Taking Control of Costs in the Cloud
After culture, people, and process, the most common topic I hear about is the surprise organisations have in shifting their thinking on financial management once on-prem applications move to the cloud. If done well this is a liberating opportunity to closely align cost with value, to engage the business at large in decisions previously abdicated to IT, and to embark on an ongoing process of cost optimisation. Here I share three fundamental shifts and some basic practices to manage this cost/value balance in the cloud.
Isn’t This Obvious?
What percentage of your company’s capital is wasted by mismanaging your cloud estate? Multiple sources put this figure at 35%. While this number might be somewhat dated, I know from experience that if you don’t change your financial and technology operating models after moving off-premise, costs can quickly mount. The discipline of making costs visible and then managing them as a business imperative rather than as an IT activity is part of the culture change necessary for newly digitised organisations. It involves close collaboration with your finance department to embed a mindset of continuous improvement. There are a few best practices you can adopt that will make your move to the cloud as seamless as possible. Together, tech and finance will be an indivisible FinTech team that strategically works to both manage costs and extract value from investments.
From One-Time Project to Ongoing Visibility
Don’t you miss the days when you could build a (expensive) data centre with a fixed amount of (guessed) capacity and then not have to pay much attention to the ongoing costs until an (expensive) equipment refresh or a (emergency) capacity upgrade was required? By far the biggest shift—and relief—in moving to the cloud is being able to ditch this mentality and adapt a more predictable model. Some basic principles I suggest are:
- Aggregate your organisation’s Amazon Web Services accounts using AWS Organizations for a single, company-wide view of spend
- Conscientiously tag every service used for cost reporting, including owner, stack type (test, prod, etc), and associated application
- Determine your internal cost allocation model
- Create and evolve actionable reporting
To determine your internal cost allocation model, you must determine whether to charge cloud and internal costs out to business functions or product teams, or just make the costs visible? There are pros and cons to both approaches. Internal chargebacks drive accountability, but due to the inclusion of internal overheads, they can be seen as a tax, causing rancour due to little perceived incentive for the charging organisation to reduce their own costs. In addition, it can be a financially burdensome activity that, in effect, just shifts money around the same organisation. Showback models are simpler to administer and create less overhead and debate, but might not drive the right ownership to lower costs. Personally, I would treat driving ownership of cost reduction as a cultural shift and use a showback approach unless the true (non-company) end customer is paying for the service.
Your reporting system can initially be as simple as a Tableau or Excel report. Over time, as your management controls and complexity increase, your reporting mechanism can evolve into a more sophisticated custom setup, or you can utilise one of the many commercial tools available. Critically, the reports need to be meaningful to the target audiences. Senior management will want to see data such as trending of cost per transactions; product teams will need more granular service information aligned to their products. The latter teams should be looking at metrics such as peak and average CPU and memory utilisation, and the utilisation of environments and snapshots to identify sources of manageable waste.
From “Optimise Once” to Iterative Optimisation
Due to the variability of spend based on changes in demand and the ability to quickly spin up (and not always spin down) new environments, another shift that has to occur is thinking of the FinTech management of the cloud as an ongoing process and opportunity to reinforce basic disciplines. This discipline includes ensuring that your architecture team considers cost a non-functional requirement and continually evaluates how new technologies such as serverless can further optimise cost. Ongoing cost optimisation relies on a broad knowledge of the cost control levers that can be pulled and the tools available that can impact on the cost/value equation.
- Set improvement targets and institute a continual improvement process of Plan-Do-Check-Act to achieve incremental improvements. Drive a discipline of automating everything as a “business as usual” activity. For instance, reducing environment setup times from days to hours or minutes not only encourages practices such as tearing down and rebuilding environments regularly to ensure they retain a known state, but also reduces the likelihood of environments being left running unnecessarily by teams who want to avoid the pain of manual builds. You might want to reinforce this behaviour with rewards designed to identify the best efforts to sensibly reduce operating costs.
- Services including AWS Instance Schedulers, AWS Systems Manager, AWS Trusted Advisor, and AWS S3 Intelligent Tiering can help automate cost- and utilisation-based decisions. AWS Budget can also alert you to spend threshold breaches.
- As technology offerings continue to evolve, and your team becomes more familiar with working in an opex-centric cloud environment, education on the investment options for your FinTech team is important. Have the team use data and trending to understand when spot instances, reserved instances, or the new savings plan can be used to optimise costs. This is a great example of a living evaluation with new functionality or increasing traffic potentially changing the option which is most cost effective for you.
- As part of your product development disciplines, ensure sufficient importance is given to non-functional requirements. With existing applications, simple changes might enable better utilisation of elastic capacity.
Jake Burns has further advice on ongoing optimisations here.
From an IT Concern to a Business Decision
While managing cloud-based costs is different to on-prem solutions and might seem daunting, it’s liberating when teams realise the control they have when using the cloud. Discussions with the CFO and CEO can evolve to include topics such as the target cost per customer transaction and associated cost drivers, the appropriate cost/reliability trade off as application usage grows, or business opportunities unleashed by new technologies. The ability to quickly adjust the capacity and configuration is a level of control that few businesses previously enjoyed. To keep this business conversation alive, it’s important a business sponsor is appointed, ideally in the C-Suite, and that they understand the trade offs they can make. While cost is important, reliability and efficiency also demonstrate business value. Find the right KPIs for your business that emphasise this. The appointment of a business sponsor is an excellent opportunity to increase the level of IT education and empowerment in an organisation. In partnership with Finance, what used to be an IT issue is now a business decision. Instead of presenting on-premise application and infrastructure costs as a given, which had very little “decisionability,” now you can engage your business partners in topics where they too can feel empowered.
Embrace your finance team, take control, and turn cost management into a strategic and interesting business discussion.