In general, IT teams want cloud resources to be up. The higher your uptime rates, the better your business can meet SLA compliance requirements.
But as Gartner analyst David Wright put it at the company's recent IT Infrastructure, Operations and Cloud Strategies Conference, there is such a thing as "unwanted uptime." In other words, there are situations where you have resources running that don't need to be.
When that happens, you end up with unnecessary cloud computing costs. That's why uptime management, as this article explains, is a critical step toward managing cloud costs. Keep reading for an overview of what IT leaders need to know about the financial dangers of running unnecessary resources in the cloud.
What is uptime, exactly?
It's common for IT professionals to use the word "uptime" in a generic sense, as if all measures of uptime are the same. But in reality, there are two fundamentally distinct types of uptime:
- Service uptime, which refers to the amount of time that services are available in a way that meets user requirements.
- Infrastructure uptime, which is the amount of time that infrastructure components, such as cloud servers, are up and running.
Infrastructure uptime drives service uptime, of course. You can't make services available if you lack sufficient infrastructure to host the applications that power those services.
Excess uptime and cloud computing costs
That doesn't mean, however, that maximizing infrastructure uptime is always ideal. In some cases, as Wright explained at the Gartner conference, the amount of infrastructure capacity that you need to meet service uptime requirements can vary widely.
For example, if you run an eCommerce business, your applications probably experience higher demand in the holiday season, and you need more infrastructure uptime during that period to ensure that you meet service uptime expectations. But when business slows down after the holidays, you can maintain the same service uptime with less infrastructure uptime.
Striking the right balance is important, of course, because too much infrastructure uptime leads to higher cloud computing costs without an increase in business value. The more servers, databases and other resources you have running in the cloud, the more you'll pay in most cases – whether or not the cloud resources are actually necessary to support your service uptime requirements at a given point in time.
This challenge is exacerbated by the fact that cloud providers typically don't tell you when your infrastructure is over-provisioned. They'll happily provide as many VM instances, storage buckets and other resources as you request, regardless of whether they're necessary.
Tips for managing cloud costs
That means it's up to you to figure out how to manage your cloud costs without sacrificing your ability to maintain service uptime requirements. Common strategies include:
- Configuring autoscaling policies, which can automatically scale cloud infrastructure footprints up or down in response to shifts in demand.
- Using cloud rightsizing tools, like AWS Compute Optimizer, to predict which types of cloud infrastructure configurations are most cost-effective for your workload requirements.
- Tagging your cloud resources. Tags alone don't guarantee you don't overspend, but tags make it easier to keep track of what you have running in the cloud, which in turn helps you to identify unnecessary or over-provisioned resources.
- Adding "cost observability" to your cloud, then comparing cloud performance metrics with cloud spending metrics. If you notice that your cloud performance levels have remained the same while your cloud spending has increased, there's a good chance that you're paying more for cloud resources than you need.
- Negotiating an enterprise services agreement with your cloud provider. These agreements – which are typically only available for larger companies with high-volume cloud consumption requirements – make it possible to obtain discounted cloud services, which reduce your overall cloud bill without compromising on service quality or availability.
These strategies make it possible to build a cost-resilient cloud by avoiding unnecessary cloud computing costs while simultaneously maintaining high levels of service uptime. The result is a cloud environment that delivers the uptime your business and customers want, without the “unwanted" uptime that needlessly drives up cloud costs.