The cloud can deliver extraordinary flexibility, scalability, and performance, which is why so many utilities are flying to the cloud. But while the cloud may offer dramatically reduced overhead compared to equivalent on-premise capabilities, it is still important to keep up with regular cloud optimization practices are essential for ensuring that your cloud continues to deliver the greatest possible ratio of value to cost.
In this blog, we cover five of the most important ways you can keep your organization’s cloud infrastructure optimized.
Why is it important to keep your cloud optimized?
Today’s cloud technologies are so flexible that many organizations can “lift and shift” their entire onsite technology stack to the cloud. While this option is valuable for many organizations, it creates a risk that the cloud environment is not tuned for the best performance and will never be truly optimized. “Migrate now and optimize later,” can quickly become “why is my cloud bill so high?”, and in our experience, establishing some best practices for optimally managing your new cloud environment will almost always pay off in the long term.
As we explore below, flexible options for procuring cloud resources can drive substantial savings with minimal operational impact. Yes, the cloud offers unprecedented operational and financial flexibility. But to generate the greatest possible business value, an organization’s cloud resources must still be actively and regularly optimized. Doing so successfully requires attention to some core best practices that will be new to organizations that are still transitioning to the cloud.
Five Essential Practices for Cloud Optimization
- Find Unused or Unattached Resources
Ensuring efficient employment of cloud resources is the foundation of any successful cloud optimization process. Locating any unused or unattached resources is a great place to start. These resources are likely to accrue over time as instances are set aside and eventually forgotten, leading to cost bloat if not weeded out. Keep in mind that cloud vendors charge based on the full set of resources an organization has access to, not just active resources, so you may be charged for instances that go entirely unused.
- Select, Provision, and Right-Size Resources
The full variety of different instances and resources available from leading cloud vendors can be truly overwhelming, and it can be easy to choose unnecessarily costly options. An important check on this potential waste, “Right Sizing” refers simply to a process for aligning instance types and sizes with the businesses’ performance and capacity requirements. This work should include both instituting a more systematic process for selecting new instances and pinpointing opportunities to eliminate or downsize instances without impacting the business.
- Reduce costs by more than 60% for Non-Production Environments
Overall uptime is included in most cloud SLAs, often with different tiers of promised uptime and associated costs. For business-critical systems, it may be important to ensure the greatest possible level of uptime, but these costs should be carefully aligned on an environment-by-environment basis. For example, non-production environments may not need 24/7 coverage, particularly if uptime can be scheduled for normal working hours.
- Invest in AWS Reserved (RIs) or Azure Reserved VM instances (RIs)
Most cloud vendors offer different purchasing options for cloud resources, such as On Demand (Pay-as-you-go), Savings Plans, and Reserved (RI) Instances. In general, On Demand resources come with a substantial cost premium (commonly 70%-90%) compared to Reserved Instances.
In short, RIs provide a substantial discount for a given VM instance type for a designated period (typically 1-3 years). The best course is to strategically identify predictable and variable workloads and combine Azure Reserved Instance rates with pay-as-you-instance pricing to manage the costs without impacting the business outcomes. This approach not only the lower total cost of ownership but improve budgeting and forecasting and lowers your upfront cash outflow with a monthly payment option at no additional cost. And if a given capability only requires an instance to be scaled up occasionally (such as Outage Management Systems for utilities in the wake of storms), a reservation discount can apply to the other sizes that are listed in that same instance size flexibility group.
- Take Advantage of Spot/Preemptible Instances
Spot and preemptible instances (different vendors use different terminology) refer to another arrangement where instances are sold at a substantial discount compared to On Demand. In this case, processes in these instances may be suspended (preempted) if the vendor needs to reclaim this computing capacity during high-demand periods.
Of course, this approach is untenable for some business-critical applications. But it can be a great source of savings for tasks that are not time-sensitive and can withstand interruptions (such as remote desktop services, and large batch processing of data).
Need help optimizing your cloud?
The cloud offers unprecedented flexibility, but we recognize that the options can be overwhelming, particularly if your organization is new to managing and optimizing cloud-based resources. We encourage you to contact our team if you need help.
HEXstream has delivered utility analytics solutions for some of the largest utilities in North America. We can support clients with not only their analytics-related challenges but implementing the best practices needed to get the most possible value out of the cloud-based infrastructure that unlocks new possibilities for analytics.