Public cloud computing is a must if your organization wants to scale quickly and meet the day-to-day, ever-changing demands of the business. One of the challenges for IT, though, is controlling cloud consumption costs while moving workloads.
A recent survey1 found that cloud over-spending was higher in 2022 than in the previous year: 56% of companies surveyed admitted that spending on public cloud was significantly over budget, some by over 20% to 30% of their intended spend.
While there are multiple factors related to unexpected spending in the cloud, some of the most common are:
Scaling resources to address unexpected demand
Lack of resource utilization governance
Failure to take advantage of cloud provider discounts
With concerns over cloud costs rising, several organizations have scaled back their cloud use, limiting their growth potential. Some organizations are even considering cloud repatriation or a move back to a data center-first approach. But there’s another option: cloud management platforms and FinOps to better manage costs.
In this blog, we’ll look at five simple ways to optimize spend with IBM Turbonomic — a hybrid cloud cost optimization platform and the “Ops” in FinOps — while still leveraging the elasticity of the cloud.
1. Proper resource utilization with rightsizing
Rightsizing is the action of matching workload types and sizes to your instance performance and capacity requirements while keeping costs down. Turbonomic analyzes application performance with its AI-based insights and monitors demand to ensure that cloud resources are right sized to meet demand. The solution continuously analyzes application performance and demand and recommends the optimal resource configuration. It does this by considering a variety of factors; first and foremost is application performance. Turbonomic monitors the applications performance and identifies any bottlenecks that are causing issues. When appropriately sizing virtual machines (VMs), Turbonomic software considers VM virtual memory, CPU, storage, IOPS and latency, as well as demand across the entire infrastructure. Also, Turbonomic looks at current resource utilization of cloud resources and identifies under and over-provisioned workloads and can take actions to properly resource each workload optimally in real time to keep performance matching demand.
2. Meet demand with Autoscaling
Autoscaling is a cloud computing feature that enables IT admins to scale cloud services such as server capacities or VMs up or down automatically based on defined policies created by the admin team. While many public cloud services offer Autoscaling service, there are limitations to customizing public cloud tools as they are designed around basic, simplified policies that need to be implemented for all scenarios.
Turbonomic offers a more intelligent approach. It uses machine learning to analyze application performance and demand and then make real-time recommendations. These immediate actions prevent application performance issues before they happen. How does Turbonomic do this?
Turbonomic’s full-stack visualization of resources across the entire infrastructure can quickly identify potential resource issues and automatically scale cloud resources up or down based on demand. This ensures that applications continue to run as needed and that organizations are only paying for the resources they need when they need them.
Turbonomic’s “scaling but not overprovisioning” is done in two ways:
Use machine learning through its AI insights to predict future demand for cloud resources. This allows Turbonomic to automatically scale resources up or down before demand spikes, which helps prevent performance problems before they happen. This helps prevent overprovisioning to protect application performance, resulting in saving wasted cloud consumption of resources.
Automatically rightsize cloud resources, such as VMs or containers based on their current utilization. It can also shut down idle instances or unused resources or move workloads to a more appropriate cloud instance that will enable the application to perform better and, in some cases, reducing long-term cost.
While other public cloud tools may offer recommendations to autoscale, most don’t take cost in effect. Or they may recommend a larger instance when dealing with performance issues. But Turbonomic helps you focus on the big picture of controlling costs while assuring application performance.
3. Managing reserve instances
Customers typically buy reserve instances (RI) to save money and maintain flexibility. With reserved instances you purchase upfront, often at a discount. Ideally, you then leverage them when there are historical times of high demand—think black Friday for retail, or when those Taylor Swift concert tickets go on sale. The challenge for IT can be when to use these RIs and when to purchase them.
Using AI-insights and embedded automation, Turbonomic leverages RIs by recommending when to use and when to buy RIs. It can also automatically purchase them when it is cost-effective to do so and save time for admins to manage.
Turbonomic uses several factors of determine when to recommend and purchase RIs, including:
Current usage of cloud resources
The cost of RIs
Predicted future demand for resources
The cost of on demand cloud resources
Turbonomic provides admins the option to purchase RI through actions when it detects that a resource is consistently being leveraged at a high rate over time. This helps ensure that organization are not overpaying for on-demand cloud resources. To build trust in the process, Turbonomic provides details within the suggested RI purchase outlining why the RI action should take place and the projected improvements. That provides admins with the insights they need to clearly understand before acting.
4. Leveraging spot instances
Spot instances are unused cloud capacity that is available at a discounted price. The challenge with spot instances is that they can be reclaimed for other workloads at any time. That means they are not ideal for workloads that require a high level of availability. Turbonomic can use spot instances to help customers save money by automatically launching and terminating spot instances based on demand.
Another example: Turbonomic can leverage spot instances for workloads that can tolerate interruptions, such as development environments or batch processing jobs. Since batch processing is often run during off hours and/or overnight, Turbonomic can launch spot instances to run the workload then terminate them when that workload is completed.
Turbonomic can also monitor spot instance usage to identify potential saving as well as forecast when these spot instances can be used based on historic demand. This safely unlocks the potential of spot instance usage without jeopardizing application performance.
5. Eliminate cloud waste through optimization
While the previous four steps can help reduce cloud waste, this last option is especially key: proper resource utilization. Too often, admins over-provision resources—especially for critical business applications—to ensure performance during times of high demand. But this leads to a waste of resources during normal business operations. This is where Turbonomic shines because it can automatically distribute workloads across available resources, shut down ones that aren’t being used and reduce waste.
Turbonomic leverages it’s AI-insights to add or reduce resources on demand to prevent application performance issues before they happen. This enables IT to run properly utilized workloads in the cloud at the lowest cost.
Explore IBM Turbonomic today
If you are searching for ways to reduce cloud waste and cost, consider Turbonomic. This powerful tool can help you realize true cloud elasticity for your business. Explore the IBM Turbonomic interactive demo to see how IBM Turbonomic works across your entire cloud and on-prem hybrid environment.