FinOps: Improving Digital Efficiency

Traditional companies have adopted public cloud platforms at a breakneck pace over the past five years. The digital transformation agenda first fueled this adoption. After all, if every company should become a technology company and digital excellence is imperative, it makes sense that cloud platforms are one of this movement's cornerstones. The pandemic provided the second boost for this public cloud adoption. With concerns about business continuity and expanding the digital touchpoints with end customers at the forefront, many companies decided to double down on public clouds. AWS, Microsoft, and Google have grown aggressively to date. Until now, that is. Fear of recession and economic slowdown have moved efficiency to the fore.

Cloud Financial Operations (FinOps) are getting prioritized on IT executives' agendas. In this article, we'll share our take on the essential elements of FinOps and how CI&T has been helping companies all around the globe, from major sports leagues and financial services institutions to CPG multinational companies, to optimize their cloud expenses.

A typical FinOps practice is composed of three major components:

INFORM

The inform phase boils down to understanding what's driving costs, properly allocating expenses, and creating efficient benchmarks. You must know where you are with respect to billing at all times on your cloud journey. It starts with a cloud expense attribution model, mapping existing cost centers and business units. All major cloud platforms allow you to tag resources (projects, folders, or individual components such as servers) and export billing data with this information. Thus, the first pulse check for a proper FinOps deployment is: Do we know precisely how much we spend per product, value stream, BU, or other relevant dimension to my business?

The natural next step is defining a chargeback or showback mechanism. We strongly believe that the best (and perhaps only) way to keep cloud expenses under control is with a chargeback process. FinOps responsibility is usually across a few areas: Finance, Infrastructure, and Product/Business. For the first two, skin in the game is a natural process. The most effective way to hold Product teams accountable for expenses is by charging them for cloud consumption. If charging back business units is not possible for your company, the bare minimum is to create a powerful showback mechanism.

Forecasting cloud expenses can also be a challenge. This is particularly true during a cloud migration process. You have many moving pieces (on-prem data center, new cloud environment, people to reskill, different practices, and technologies) and little experience with the new architectures. Defining a budget for each cost center is critical, even if it's just a line in the sand at the start. It's all about accountability and creating a culture - sooner rather than later - that cloud expenses matter. One way to drive this discussion is to leverage the concept of Unit of Economics - a way to analyze a company's cost-to-revenue ratio relative to its basic unit. For instance, for an e-commerce website, how much money are you spending per order placed? Does it make sense? This is a great way to promote business-oriented conversations and adjust your cloud investments based on outcomes. Over time, with more historical data, established architectural patterns, and refining your forecast approach, the budget definition will become more accessible and accurate.

OPTIMIZE

The optimization phase deals with the actual implementation of improvements. Once you put the "Inform" engine to work, you can detect anomalies and identify initiatives that need attention and action. We split optimization tasks into five buckets. Here are the possible approaches from the least to most intrusive:

1. Commercial Agreements and Committed Use Discounts (CUDs): You can drastically reduce monthly invoices by implementing more favorable commercial agreements and deep commitments with the cloud provider. The tricky part is planning the volume (e.g., the number of instances) and contract term (usually one to three years). The billing data export and forecast exercise mentioned above are critical for this.
   
2. Unused Resources: Identify and remove resources the company is not using. Over time, it's common to find compute instances, persistent disks, and other unused resources generating costs with no associated value.
   
3. Rightsizing: Historically, data centers use only 10-20% of their computing power. Many companies replicate this pattern in the cloud. However, with rolling updates, zero downtime upgrades, and auto-scaling, there is no need to waste money with oversized servers.
   
4. Dynamic Provisioning: One of the cloud's most remarkable benefits is scaling on demand. New resources can be spun up to accommodate seasonalities and utilization peaks. An architecture that supports this capability and a proper configuration/setup is required, but it's worth deploying for the mid/long run.
   
5. Re-platform or Re-architect products: getting rid of expensive middleware, redesigning the architecture for auto-scaling, and leveraging optimized managed services for specific tasks can go a long way in reducing cloud expenses and improving customer experience. 

OPERATE

The operate stage defines and reviews internal processes and operating models to keep costs under control. This is likely the trickiest part because it touches your company's cultural aspects and organizational structures. As mentioned before, FinOps is a multi-departmental initiative. Why should people care about it? How will you promote engagement and accountability?

Chargeback is part of the answer. We've seen companies attach cloud expense targets to variable compensation. Other organizations rely on collaboration and knowledge sharing to achieve their goals. There is no definitive solution for this challenge, and each scenario requires discussion and customization. That said, here are some standard best practices that we see with some of our customers:

• Monitoring and alerts: Leverage built-in monitoring/alert features from your cloud provider or third-party tools like Cloudability (from Apptio) to do the job.
• FinOps Team: Define a team of people with clear responsibilities to keep the practices running and the optimizations prioritized.
• Cross-team PDCA (Plan-Do-Check-Act): Facilitate a recurring meeting with key stakeholders to discuss the practice's current situation, learnings, and future evolutions.
• FinOps Book of knowledge: Consolidate best practices, lessons learned, and patterns/anti-patterns to proactively design cost-effective solutions and quickly fix problems when something derails.
• Shared goals: If business, infrastructure, and finance share the same goals, it's much easier to align priorities.

FINAL CONSIDERATIONS

FinOps is essential and can help your company operate much more efficiently. The sooner you deploy FinOps, the better, particularly from a cultural standpoint. It's never too late to start. The good news is that it's usually a "self-funded" initiative. Depending on your current spending, you can save 5-10x more in cloud expenses than the money invested in deploying the practice.

CI&T can help you with the following FinOps priorities:

• INFORM: Billing export, billing visualization, cost-center structure, chargeback/showback mechanisms, tagging cloud resources (using Terraform or similar), budget/forecast definition, and more.
• OPTIMIZE: Define your CUDs strategy, automatically detect unused resources, identify rightsizing opportunities, and re-platform/re-architect existing solutions to support auto-scaling and cloud-native architectures. We can also augment your product teams to implement FinOps epics and stories.
• OPERATE: Implement and deploy an effective monitoring/alert mechanism, orchestrate PDCA meetings across different teams, build/maintain a FinOps knowledge base, and ensure that your FinOps practice is running seamlessly end-to-end.

---