Inflation will have an impact on your cloud costs. If you are not talking to your CFO about that impact, you should schedule that meeting for tomorrow.
Businesses of all shapes and sizes are facing significant headwinds. Costs of borrowing and inflation are forcing organizations to deliver the same performance and quality outcomes but in a cost-efficient manner. Budgets are shrinking and every department in an Enterprise is being asked to do more with less. IT organizations are not immune to this trend. Beyond acknowledging trade-offs and compromises made along the way, there’s a fundamental revisiting of whether the value of the cloud proposition was oversold.
Authors: This is a collaboration between Arka, Chris V, Houston and Jamil.
“How you locate your workloads as your needs evolve will dictate your solvency” Arka
“The market can stay irrational longer than you can stay solvent” Vernooy
“All technology is transitional technology.” Houston
“Cloud is the new Mainframe” Anonymous
Inflation and Cloud Costs
As the money supply tightens and inflation continues to rise, organizations are reconsidering their cloud costs and seeking to be even more deliberate about how they manage that spending. Recently we did a back-of-the-napkin calculation for a generic workload running in the cloud based on the government inflation calculator. For every $1,000 you currently spend, you’ll likely have to spend $1250 a year from now for the same item or said another way, $100 of spend a year or more ago only has $75 worth of compute now, based on current trends.
10 Cloud Considerations
Recent announcements from cloud providers bear this out. With that price volatility, if you’re wondering whether the cloud delivers as promised, you’re not alone. So we created a list of 10 items that organizations should consider while finding a path forward – whether that’s in rehoming on-premise, expanding in the cloud, or any point in between.
- Learning the language of your Finance department and CFO is critical. Terms like OPEX and CapEX (Operational and Capital Expenditures) should be part of your professional lexicon. Purchasing upfront timeframes of cloud resources can fall under CapEx, but that takes just as much if not more planning than traditional environments. You’re essentially committing to re-enter the world of “bounded compute and storage” which is one of the constraints the cloud originally aimed to alleviate. Similarly, the running costs such as networks, VPN, S3, and other monthly cloud resources are still under OpEx. Understanding the finance landscape is key to defining your technology choices and protecting your company’s investments.
- There are many hidden costs associated with cloud “Lift +Shift” migrations. Outsized storage costs and inefficient use of compute resources are just a start. It’s lucrative for consultancies that move on-premise systems to the cloud, but hasn’t been great for organizations that are now facing escalating AWS, GCP, or Azure bills. Time and business pressures mean working groups will skip over revisiting data policies and application performance. However, engineering teams that leverage refactoring and re-engineering in-place before migration will not only make their work life better but it will also serve to constrain future cloud costs.
- When building in or migrating to the cloud, have an asset tagging strategy and implement it consistently. Embed resource tagging deep into application configuration mapping. Without it, attributing costs can be very difficult and makes preparation for necessary finance and technology conversations another hidden cloud expense.
- Not everything can be moved to the cloud. And yet there’s a stigma attached to non-cloud solutions, and both should be addressed. Data residency and compliance are among the issues that can prevent cloud from being a complete solution, but there are other factors that can be equally valid. It’s important to maintain modern non-cloud solutions as viable alternatives and reassess your organization’s specific needs as regulatory compliance and other factors of the technology landscape continue to change.
- It’s a common mistake to over-build a cloud solution simply because it’s possible. A common example has been the popularity of splitting monolithic applications into web-based microservices. Despite its prominence, this pattern can create its own set of challenges around observability, running costs, troubleshooting, or even team organization. Once you finally see the downsides for what they are it’s often too late, and the “sunk cost fallacy” takes hold. By contrast, a monolithic application can make essential business complexities easier to manage, as coupling and cohesion match the system design and application goals. One of our guiding principles is “the answer to complexity is clarity, *not* simplicity,” and along with that, sometimes the most important decisions are the ones you choose to not implement.
- “Make it work. Make it readable. And then, if it matters, make it fast.” It’s important to acknowledge organizational pressures will force decisions you may not be able to immediately return to remediate. We encourage creating a well-defined “Definition of Done” for the workloads you’re moving to or building in the cloud. Alignment on this will be helpful. As we outline here, performance *always* matters. Support honesty about the cost-to-value ratio of the workloads that teams create.
- It’s important to talk through technology choices as they impact all stages of a project. Languages like Python are increasingly adopted for their “ease of entry” (check out Github’s state of the Octoverse). However, they can create their own set of hidden costs that other tech stacks may not incur. Solving a small-scale problem with an ‘easy’ tool may lead to early wins, only to run out of time and/or money before it’s production-ready with full-scale storage and compute. The “easier” choice is not always the better choice.
- If you opt to return to on-premise, be sure you’re not trading evil for evil – primarily overlooking opportunities to optimize on the way back (from the cloud). This will require some candid discussion and honest negotiation around how to create an efficient path forward (see point 2).
- If you’re considering your own data center, be certain to account for all of the costs. Data center management includes not only “racks and stacks”, but also the cost of physical facilities, climate control, fire suppression, multiple power and data trunk line routes, backup generators and related fuel costs, the list goes on. It all contributes to expenses and you might never see them itemized in your cloud bills, but those costs are definitely part of the picture. If you plan to go your own way, plan carefully.
- Hybrid models (a mix of on-prem and cloud workloads) are here to stay and candor about what that means for an organization is key. The “all or nothing” posture isn’t one that teams can commit to without serious consequences. While the rising costs are the primary factor everyone is seeing today, there will always be another advance, another regulatory change, or another business requirement that will shift the landscape again. While the hybrid model can be the most demanding of a technology organization, it also provides the most options in an ever-changing world.
Conclusion
We see teams with fresh ideas and bold plans get cast about due to market forces. It’s important to have a healthy dialogue within your organization, especially your CFO, along with business partners about those forces and considerations. Not everything is “cut and dried” and it usually comes down to how you frame trade-offs and costs which determine your measures for success. While macro-economic conditions may have an outsized influence on those conversations now, the flexibility gained will eventually lead to a more resilient organization that can calmly face change in any economic weather.