Thinking Like A CFO At The Network Edge

Article originally published on Forbes.com. Image courtesy of Getty.

When technology leaders discuss the value of edge infrastructure with others on their executive leadership team, the focus of the conversation is typically, and understandably, centered on the technology itself. But the edge only makes sense when a solid business case can be built for it among key stakeholders across a company. Financial design is as important as technical design. Companies that build sustainable, successful edge networks don’t need to have the most locations. They need to prove edge unit economics.

There’s always a risk of leaders treating edge like a shiny new technology project, when it is actually more like an investment portfolio. Diversification diffuses risk, which is why I advise companies to build edge ecosystems that are open, disaggregated and interoperable. And investments should be tied to measurable goals. In this case, edge infrastructure is a portfolio of small business bets, each one linked to concrete outcomes, such as improved latency or customer experience.

Designing Edge Infrastructure With A Business Perspective

You have to be able to explain to your CFO why placing your infrastructure closer to your end users will boost revenue or reduce cloud costs. If you can't defend it as a business decision, then you're not doing edge computing correctly; you're just building expensive remote data centers. I recommend building edge infrastructure with the following three principles in mind.

1. Start With A Business Constraint, Not A Technology Target

Identify a specific business constraint that you are trying to resolve. What is scarce or expensive in your business today? For example, are you concerned about the cost of cloud egress, backhaul or SLA penalties? Outline the details of an edge solution, and see if it really makes sense for your business right now. There are times when you dig deeper into a challenge and discover that the solution is not what you initially imagined.

2. Model Economics At Workload And Regional Levels

Think about an edge infrastructure project like a P&L statement, calculating profit and loss for your company’s edge footprint at the workload and the regional level. Who is paying for each line item? What is driving utilization? What's the payback cost period for each cluster or each point of presence (PoP) that you're deploying? Is each one profitable?

My company supports customers that do business all over the world and have many different reasons for building edge infrastructure. My team and I look at each customer’s project on a per-location basis, examining where their customers are and what the P&L is for each place, as well as other strategic factors. In fact, it may make sense to add a location that is not profitable initially when there is another compelling business reason at play.

Be intentional about building cost levers into your architecture. If you can deploy multiple customers or workloads across the same hardware, you can spread out and amortize costs and maximize revenue. One thing many people don’t know about building at the edge is that you don't have to burst “in region.” In other words, you can temporarily support traffic spikes with burst capacity from other strategic edge locations.

A strategy we use for larger customers is building "super PoPs" for more cost-efficient burst capacity. Maybe you only have three super PoPs in the U.S., in New York, Chicago and L.A., and 20 other PoPs in other large cities, including Phoenix, Dallas, Miami and Philadelphia. If you need to burst traffic in Miami, for example, you can do so from your New York super PoP instead of taking on the expense of building and operating another location in Miami. Smart routing techniques, such as anycast technology and traffic steering, can help you serve regions effectively from the nearest and most economical PoP.

3. Use Flexible, Shared Infrastructure

Design your business models and your edge architecture together. Calculate the total cost of your infrastructure project, including both operating expenses (OpEx) and capital expenses (CapEx). There’s the cost to acquire, install and operate hardware, but there is also the cost to decommission it when it’s no longer usable. Where you place workloads can also change key factors, from overall risk to depreciation schedules. Do the math on a per-region basis, and see if owning hardware or adopting a shared, flexible model is better from both a business and an engineering perspective.

Inflexible, proprietary edge investments become a graveyard of stranded capital. I recently received an email from the leader of a large global company that is trying to sell a large number of obsolete servers. The company can’t use the servers because they don’t meet its current requirements, yet it can’t just throw them away. It has to abide by legal, insurance and data management regulations. In the meantime, it has to keep paying for real estate and energy costs to store unused hardware. We work with customers to set up flexible, shared infrastructure that they can use for different customers and services, enabling them to expand into or withdraw from different regions without the up-front investment of buying hardware.

Look beyond OpEx and ROI, and connect every infrastructure decision directly to CapEx. Treat every region as an investment with a target payback period. Understand the short-term and long-term factors that can affect your investment, such as currency conversions, exchange rates, taxes and liability. Who is going to pay for this? What are the contract terms? What’s the break-even period?

Clearly define what you are trying to achieve, whether that’s reducing cloud egress by 30% for a certain product or improving conversion rates by 5%. Use the edge to unlock new revenue streams and cost savings, and avoid unique snowflake configurations that will require more care and feeding than your business can support.

I love the technical side of our business; I see it as the fun part of what we do. But our business, as well as yours, and those of your customers, need to start edge infrastructure planning with financial objectives for the company as a whole top of mind. When it comes to edge computing, the best business cases are as distributed as the infrastructure.

‍