Cloud computing has many benefits. But there are also challenges, and cloud cost management may be one of the biggest. Here are 49 stats, benefits, and hard truths you need to know about cloud cost management, along with tips and requirements to help you take control and keep your spending in check while delivering on all the value you’re looking for.
14 Important Stats About Cloud Cost Management
1. Worldwide end-user spending on public cloud services is forecast to grow 18.4% in 2021, with the cloud projected to make up 14.2% of the total global enterprise IT spending market in 2024, up from 9.1% in 2020. (Source: Gartner, Gartner Forecasts Worldwide Public Cloud End-User Spending to Grow 18% in 2021, November 2020) Why it matters: The cloud is a growing portion of technology budgets, so cloud cost management has a growing impact on those budgets.
2. 88% of enterprises have placed more than one-quarter of their workloads in the public cloud, and 44% are running more than half of their workloads in the public cloud. (Source: Virtana, State of Hybrid Cloud and FinOps, September 2021) Why it matters: Nearly every company is already in the cloud, so cloud cost management is a challenge that most must address.
3. 84% of companies are taking a multi-cloud approach with workloads deployed in more than one public cloud. (Source: Virtana, State of Hybrid Cloud and FinOps, September 2021) Why it matters: Nearly every company is already using a multi-cloud approach, so cloud cost management across multiple public clouds, with all the added complexity this entails, is a challenge that most must address.
4. FinOps is the practice of bringing financial accountability to the variable spend model of cloud, but only 15% of enterprises report having a mature and evolving FinOps practice. (Source: FinOps.org, The State of FinOps Report, 2021) Why it matters: Most organizations are still a long way off from operationalizing financial accountability of cloud spend.
5. 72% of organizations reported having to repatriate applications after migrating them to the cloud. Of those, 20% said they had to roll back because of unexpected costs in the public cloud. (Source: Virtana, State of Hybrid Cloud and Migration, February 2021) Why it matters: Unanticipated costs can be severe enough to require companies to hit “undo” on their (often mission-critical) cloud migration projects.
6. Only 3% of enterprises surveyed say they don’t have any wasted cloud resources. (Source: Virtana, State of Hybrid Cloud and Migration, February 2021) Why it matters: The chances that your organization doesn’t have any wasted cloud resources—and therefore wasted spend—is practically nil.
7. One-third of cloud spend is wasted. (Source: A variety of online sources indicate that the range is between 30-35%, which tallies with our own experience—the average wasted public cloud spend we’ve uncovered among our customers is 32%). Why it matters: Consider the size of your cloud budget; one-third of that number is probably a lot of money—wouldn’t you rather have it available for other projects?
8. 82% of organizations with workloads running in the public cloud have incurred unnecessary cloud costs. Reasons cited include workloads bursting above agreed capacity, overprovisioning of compute or storage resources, overbuying or having unused reserved instances, storage blocks that are no longer attached to a compute instance, and poor job scheduling. (Source: Virtana, State of Hybrid Cloud and FinOps, September 2021) Why it matters: There’s no one reason for out-of-control costs, and no one simple fix for better cloud cost management.
9. 81% of survey respondents say that surges in cloud utilization and costs have an impact on their team. (Source: Virtana, State of Hybrid Cloud and FinOps, September 2021) Why it matters: It’s not just a budgetary challenge, it has a human cost as well.
10. Only 36% of enterprises have comprehensive, unified visibility and management capabilities across all public clouds. (Source: Virtana, State of Hybrid Cloud and FinOps, September 2021) Why it matters: Most organizations are trying to manage their cloud costs with inconsistent data and blind spots.
11. 62% say that getting a global view of cloud costs requires the use of a lot of different systems/tools/programs. (Source: Virtana, State of Hybrid Cloud and FinOps, September 2021) Why it matters: The more time you spend stitching together data, the less time you have to actually analyze that data and then do something about it.
12. 72% of cloud decision makers are fed up with piecing together data from multiple IT operations tools. (Source: Virtana, State of Hybrid Cloud and FinOps, September 2021) Why it matters: In addition to point #11 above, it’s also frustrating your team and potentially contributing to burnout and expensive turnover.
13. 40% say it takes days, weeks, or even months to get an up-to-date global view of cloud costs. (Source: Virtana, State of Hybrid Cloud and FinOps, September 2021) Why it matters: The longer it takes to get an “up-to-date” global view of your cloud costs, the less likely that information is, in fact, up to date.
14. 27% of cloud teams rely exclusively on periodic audits to learn about cost-impacting changes, and 26% get that information only from their end-of-month bill. (Source: Virtana, State of Hybrid Cloud and FinOps, September 2021) Why it matters: The longer it takes to spot changes that drive up your costs, the more you’ll spend before being able to make any needed adjustments.
3 Cost Benefits of the Cloud
15. Shift from capex to opex. Because the bulk of the physical infrastructure and many application costs are assumed by the cloud provider, organizations can reduce their capital outlay in favor of an operating expense model. The extent to which you can make that shift will depend on the type of deployment (SaaS, PaaS, IaaS, etc.), but you will save on up-front capital equipment and software, which can result in net positive cash savings.
16. Pay as you go. On-premises deployments are often sized to support the maximum load required, even if that peak is only reached for short durations (think Black Friday for a retailer or end of quarter for financial accounting), which means that resources may sit idle 90%+ of the time. In the cloud, however, you’re only using—and should be paying for—resources as they are needed, which can result in substantial cost savings.
17. Reserved instances can deliver additional savings. If you commit to a specific capacity and resource allocation to be used over a certain time frame, you can get a reduced rate.
14 Hard Truths About Cloud Cost Management
18. Simply moving to the cloud doesn’t guarantee lower costs. IT Infrastructures are complex with a lot of variables that need to be taken into account. The decisions you make at any point in the process can potentially have short- and long-term impact on your cloud costs.
19. Keeping the traditional data center mindset in the cloud will cost you big time. Many engineers steeped in the data center capex approach simply aren’t used to having to account for cloud costs and they haven’t built up that muscle. But if they buy cloud resources based on expected future growth and keep services running all the time irrespective of actual utilization, the way they do for on-premises deployments, the organization will rack up big bills for capacity they don’t actually need.
20. The array of cloud configuration options is overwhelming. The total possible configuration permutations number in the hundreds of thousands. As if that wasn’t challenging enough, that long list is constantly changing as cloud service providers (CSPs) introduce new types, sizes, and generations on an ongoing basis.
21. Cloud vendor lock-in could create hidden costs. If your CSP’s offering is highly customized for you or if you’re using APIs that are only available on their platform, particularly if you have a lot of data within that environment, you may not be able to transition to a different platform without incurring significant costs.
22. High transaction speeds in your back-end data layer can require expensive investments. Migrating an entire application to the cloud can affect transaction speed and will likely require substantial investment to re-architect and re-write it for “all-in-cloud” operation.
23. Poorly planned reservations can cost you more money in the long run. If you lock yourself into buying capacity that you won’t use, those savings may not materialize. And while you might have some flexibility to modify or exchange a reserved instance, you typically can’t cancel it.
24. Cloud bills are notoriously complex. They can be vague and confusing, and even if they are detailed, they’re probably not categorized in a way that aligns with how you manage your business. And there can be high-dollar-amount line items that don’t tell you what’s in them—see point #25 below.
25. Some of your biggest cloud cost line items can be a mystery. The AWS EC2 Other cost bucket is a prime example. This is a catch-all for a variety of service-related usage types. The problem is that EC2 Other can show up on your bill as one of the top items in terms of dollar amount, but AWS native tools don’t let you analyze those costs in detail so it’s hard to know exactly what’s in there.
26. Reservation charges change the cost scale. When you purchase EC2 reservations, for example, your costs are recorded at the time they are charged and will appear as a single charge on that day. Because it’s a large number relative to your average daily charges, this distorts the scale of the cost axis and you lose the resolution needed to see trends, even meaningful spikes and dips.
27. “Pay as you go” goes both ways. The opex model of the cloud means that you don’t have to pay for capacity that you don’t use (see point #16 above). However, it also means that you are on the hook for what you do use, and if you’re not managing your capacity well, you could end up spending much more than you anticipated.
28. Spinning up cloud resources is easy, but so is forgetting to spin them down when you no longer need them. This matters because you could still be paying for them. For example, when you terminate an instance, depending on your settings, the attached storage volume may not be automatically terminated and you’ll continue to be charged for it. When you start automating for scale, this problem can grow exponentially.
29. Cloud forecasting is more art than science. Planning is complex and the reality is that many tools fall short on giving you the full scope of information you need in an easy-to-use way.
30. Cloud optimization guidance is often limited to CPU utilization. CPU utilization is an important factor, but there are other computing dimensions that affect performance. Memory usage, IOPS, and network bandwidth, for example, must also be taken into account when making rightsizing decisions.
31. Average workload utilization data can be highly deceptive. Rightsizing a workload requires you to understand its behavior over time, during peak business hours. Some tools will aggregate and/or sample the data to provide an “average,” but this can be problematic because it’s missing critical detail about when it hits 100% CPU utilization over the course of the time frame, how long it’s running at or near 100% during that period, and where it’s running the rest of the time (at 50%? 10%?).
5 Tips to Improve Your Cloud Cost Management
32. Balance cost reduction and risk. When it comes to selecting a specific cloud configuration, you’ve got lots of options. Like any purchase, you want to evaluate your choices based on value, not just pick the one that offers the lowest cost. That value judgment should reflect your risk tolerance. For example, let’s say option A is cheaper but if demand increases, you could run into performance issues that will affect customer satisfaction. Option B is somewhat more expensive but gives you much more capacity for growth. There’s no one right answer, but you do want to understand the implications of both and then pick the one that’s better aligned to your organization’s priorities.
33. Scale mission-critical applications without breaking the bank. As demand grows for core applications your enterprise depends on, consider moving components that can scale horizontally to the cloud, such as web or application servers in a typical three-tier web application, and maintain core components for data and transaction processing on-premises. This type of hybrid-cloud model is especially applicable if ultra-fast is mission critical.
34. Measure actual—not average or sampled—workload utilization over a representative time period. Whether that’s two weeks or two months or some other time frame will depend on your business.
35. Align sizing to workload objective. It’s probably worth spending a little more to assure capacity for a mission-critical workload with no tolerance for latency. A back-office workload that runs a process off hours and can take longer without having an impact on the business, however, doesn’t need the same capacity.
36. Use cost-allocation tags. You need to be able to parse your public cloud expenditures by individual business entity, such as department, business unit, or customer, particularly if your enterprise uses a chargeback model. You can set up cost-allocation tags to customize the way you track your resources.
13 Requirements for Smarter Cloud Cost Management
37. Balance cloud cost/performance trade-offs right from the start. Before you migrate your first workload, you’ll want to verify that it can meet or exceed performance requirements in a cost-effective manner. What to look for: The ability to group workloads with similar utilization levels and other characteristics and then iteratively model them in your target cloud environment to find the configuration that meets your performance criteria at the lowest cost.
38. Easily see all your cloud costs at a glance. If your cost data is stuck in native tools, you’ll never have more than a siloed view, making it hard to manage your entire cloud estate holistically. And if you extract and compile it manually, you create a lot of extra work. What to look for: A hybrid-/multi-cloud cost tool that will easily give you a unified global view of all your cloud costs.
39. Know what’s actually happening in your environment as it’s happening. If you wait until you get your end-of-month bill to catch spikes in spend, or for users to complain to find performance bottlenecks, it’s too late. What to look for: Real-time monitoring to detect and alert you to large shifts in performance and capacity as they happen.
40. Slice and dice your cloud cost data. To get a handle on your cloud costs—so that you’re truly managing them, not just paying the bills—you need to be able to easily perform multi-dimensional analysis. What to look for: Detailed bill analysis that enables you to group the data in a meaningful way, such as by period, attribute, or service; sort the data in different ways based on the criteria you need to assess, whether that’s service, total cost, or cost delta; and filter the data set so you can zero in based on services, attributes (such as instance type or entity name), or tags (such as owner, project, or application).
41. Create and schedule custom reports for automatic delivery. Canned reports are easy, but they probably won’t give you the information your team actually requires. What to look for: The ability to create customized reports that tell you what you need to know with easy set up.
42. Configure custom alerts. Just like reports, you need notifications that reflect what’s important to you. Otherwise, it’s just noise (or worse, silence when a warning is justified). What to look for: The ability to set up notifications to signal conditions that break the thresholds that you care about.
43. Easily find unused or idle resources that are unnecessarily costing you money. Finding and eliminating these hard-to-spot budget busters can save you a lot of money. What to look for: The ability to identify unused compute instances, storage blocks that are attached to a stopped instance or that are unattached, unattached load balancers, and idle elastic IP addresses.
44. Plan reserved instance purchases more effectively. Avoid overcommitting by finding your ideal resource settings before locking yourself in. Smarter reservation planning ensures that you can strike the optimal balance between deep savings and infrastructure agility. What to look for: The ability to accurately account for peak and non-peak usage and perform what-if analysis for potential savings in varied scenarios.
45. Quickly adjust to changes in CSP offerings. CSPs are constantly evolving and expanding the array of choices. What to look for: The ability to compare your instance’s attributes and utilization, along with any additional constraints you want to factor in (e.g., CPU utilization should not exceed a particular level), against the most current cloud SKU library and get cost-saving recommendations that support your workload requirements.
46. Enable executive- and admin-level dashboards with configurable views. The cloud team needs operational data for day-to-day cloud management, but the CFO doesn’t care about those details as long as investments are smart and budgets are adhered to. What to look for: The ability to build different types of dashboards to deliver the “right” information to different audiences.
47. Dig into EC2 Other costs. When your CFO asks what the heck this big “Other” charge is, you need to have a satisfactory response. What to look for: The ability to break down the spend in this or any other block.
48. View amortized costs. If reservation charges are distorting your cost scale, you can’t get a handle on trends in other cost categories. What to look for: An amortized view which distributes up-front fees and monthly installments evenly across the reservation period.
49. View blended and unblended costs. If you’ve consolidated multiple accounts into a single billing account and have reserved instances that are shared across those accounts, you may want different views into those costs. What to look for: The ability to see those costs blended (averaged rates of the reserved and on-demand instances that are used by member accounts) or unblended (sum of the applicable rates multiplied by the usage for each resource to represent the usage costs on the day they are charged to you).
Want more? Check out our Cloud Cost Optimization Guide.
Virtana: Your Cloud Cost Management Partner
With Virtana Cloud Cost Management, you can optimize your capacity and cost in real time on an ongoing basis. Our real-time data collection and analytics identify unused resources that can be eliminated so you can stay on budget, even as conditions and options change, and avoid an end-of-month billing surprise. With Virtana, you’ll have all the cloud cost management capabilities you need. Try it for free
Jaime Banuelos
