Chances are your organization now spends a significant amount of money on cloud resources. The question is: Are you maximizing the value of that spend?
It may seem that on-demand cloud instances are extremely cost-effective, giving developers and IT teams easy and quick access to needed resources. However, Flexera’s RightScale team has measured cloud spend among enterprise users and identified that, on average, 35 percent is wasted. That wasted cloud spend adds up to more than $15 billion across just the top three cloud providers – AWS, Azure and Google!
It’s probably easy to understand that the decentralization of cloud use and oversight can lead to wasted cloud spend. But that’s not the only cause. Let’s take a look at some additional factors that lead to this significant amount of waste:
- Cloud pricing is actually complex. The on-demand nature of cloud resources allows developers and IT teams to quickly ramp up or scale down when needed. But the flexibility comes at a cost that needs to be monitored on an ongoing basis. While cloud pricing may seem simple, the reality is there is a dizzying array of ways providers charge for access to cloud resources. Access to virtual machines and instances can be priced differently depending on the region, hours that they run, and countless other circumstances under which the resources are accessed. Pricing for storage capacity can be just as complicated with many different tiers and classes.
- Instance sizes are often overprovisioned. There are many unknowns when migrating instances from on-premise to the cloud. What is the equivalent instance size in the cloud? Are performance characteristics the same? What instances and sizes make the most sense to realize maximum value from the cloud? Unfortunately, a lot of enterprises take educated guesses and end up overprovisioned. And once instances are in place, they rarely get downsized.
- Individual resource owners don’t see the full picture. With today’s agile development processes, teams may be automatically provisioning and tearing down instances for development and QA without realizing the total costs of the deployments. Hourly costs for cloud use may seem low, but when the resources are used for weeks, months or even years, the costs can really add up. Users may not be aware that bulk use discounts are even available. But the worst part is that resource owners may not receive the reports that show the true costs of their cloud use to be able to actually understand and adjust the cloud provisioning.
- Automation isn’t being used to optimize workloads. Many organizations take the stance that one-time optimization of an enterprise’s cloud spend is adequate. But that is not the answer. Identifying waste and resolving it must be ongoing to ensure cloud spend is under control. Automation makes it much easier to optimize cloud spend by resolving issues with unused, idle or underutilized accounts; instances running in higher-cost regions when they could be in a lower-cost region; older instances that should be updated to newer versions; and more.
Taking Control of Cloud Spend
According to data from Flexera’s RightScale team, 39 percent of cloud instance spend is on VMs that are running at below 40 percent of CPU and memory utilization, with the majority of those running at under 20 percent. It’s clear that a lot of enterprises are spending significantly more than they need to on cloud resources. How can you help your enterprise reduce that waste? Start with these recommendations:
Recommendations for Reducing Wasted Cloud Spend
| Instances | Action |
|---|---|
| Idle instances | Eliminate instances that are no longer being used, such as temporary instances for projects that have ended (dev, test, demo, training, experiments). |
| Underutilized instances | Downsize instances that have low utilization of CPU or memory or switch them to a lower cost instance family. |
| Part-time instances | Schedule instances (such as development) that are only used part of the time to shut down during evenings or weekends . |
| Superseded instance families | Switch instances from older instance families to the newer, lower-cost replacement families. |
| Higher-cost regions | Move instances that are running in higher-cost regions to run in nearby lower-cost regions. |
| Storage | |
| Unattached volumes | Delete storage volumes that are no longer attached to instances. |
| Old snapshots | Remove snapshots that are beyond your snapshot retention policy. |
| Overprovisioned storage class | Downgrade storage that has been provisioned as SSD but could be HDD, or storage that is provisioned as a higher class (hot, warm, cool, cold) than is needed. |
| Other | |
| Unused services | Remove services that have been left running but are no longer being used. |
| Unused accounts | Decommission accounts where services have been left running but the accounts are no longer being used. |
| Discounts | |
| Not using discount options | Take advantage of discounting options such as reservations or other volume commitments. |
| Low coverage with discounts | Purchase enough coverage with discounts. |
| Underutilized discounts | Match the appropriate resources to already purchased discounts. |
You can do it the easy way, or the hard way …
With the right tools, cloud spend optimization can be an efficient, ongoing process. Flexera’s RightScale Optima optimization platform is the right tool and the easy way to optimize your cloud usage to realize significant savings on cloud spend. In collaboration with your organization’s cloud optimization goals, Optima analyzes your cloud utilization and automates actions to shut down unused resources or make adjustments to take advantage of lower prices or discounts. Learn more >>