Gartner expects 51% of the enterprise IT spending across core segments, like application software, system infrastructure, etc., to be dedicated to the public cloud by 2025 — up from 41% reported in 2022.
The opportunity to leverage on-demand, scalable cloud models for cost-effective business operations is encouraging organizations to expedite their digital business transformation initiatives.
However, the spending on cloud services is not without its risks, as organizations can waste millions of dollars on the cloud through a combination of committing to the wrong cloud deployment model, choosing the wrong providers, stifling cloud agility, and failing to optimize their spending.
In this article, we’ll explore some of the mistakes that IT companies must avoid when optimizing their cloud costs.
- Failing to Monitor Indirect Cloud-Related Expenses
Often, when migrating to or growing their cloud ecosystem, organizations tend to overlook indirect cloud costs. What exactly are these? In short, these are the costs that can’t be directly attributed to a product or a certain operation. Shared storage costs, networking infrastructure costs, taxes, time delays, and security vulnerabilities — all qualify as indirect costs.
As the number of users and business requirements proliferates, organizations must track and estimate indirect expenditures in advance. But that seems to take a back seat. Usually, IT departments excel at tracking technical indicators such as workload, CPU, and memory utilization. But they aren’t always equipped to analyze indirect spending indicators and assess whether the overall performance of cloud services is worth the expense.
In line with this, cloud migration costs can also be difficult to predict, and it is easy to miss expenses such as staffing and introducing new types of services. This only serves to hinder cloud cost optimization efforts.
- Not Right-Sizing the Workloads
Under-provisioning and over-provisioning cloud resources can be equally detrimental to a business. With under-provisioning, applications will suffer from performance degradation. On the other hand, over-provisioning means paying for resources and services that are not needed. Because of the dynamic cloud ecosystem, it’s exceptionally daunting to get the right workload size balance.
But, dealing with an excessive amount of unused compute and storage resources is critical to ward off unnecessary spending. It’s here that organizations should consult a cloud technology expert that can help establish a concrete (and efficient) cloud-usage model — customized to the capabilities of the specific cloud platform.
- Mismanaging Snapshots
Cloud snapshots allow for an eye-popping amount of data to be retained and easily restored to a new disk. But there is a caveat — the snapshots created on different cloud platforms (such as AWS and Azure) incur storage charges. In AWS, for example, automated snapshots (when enabled) are retained for 35 days. Manual snapshots, on the contrary, don’t expire.
Of course, these snapshots are critical for enterprises because they prove viable for disaster recovery, testing and development, data migration, etc. So, it’s necessary that organizations delete the snapshots that are no longer of use. They must also replace volumes with snapshots and remove unattached EBS volumes.
- Not Addressing Cloud Sprawl
“Enterprises often find themselves mired in an intractable sprawl of cloud services with inadequate visibility into the corresponding spend,” outline analysts at McKinsey. Naturally, this leads to an inflated cloud-related bill, which is often the result of organizations being unable to streamline and optimize their cloud services.
In essence, cloud sprawl could take various shapes. Consider this; different departments within an organization might leverage the same types of services from different CSPs. This prevents unified monitoring of the overall cloud spending and utilization. What’s more, it creates a scattered cloud ecosystem in which each department has a different set of cloud providers. In fact, such a scattered ecosystem prevents businesses from leveraging discounts that CSPs offer on their services.
- Thinking and Operating as per the Capital-Expenditure Model
We all know that cloud computing has ushered in a breed of IT service providers capable of scaling infrastructure as and when demand arises. However, when migrating to the cloud, organizations often make the mistake of long-term demand planning. This is because they are accustomed to the way enterprises have traditionally handled their physical assets based on capital expenditure. The viable alternative is to treat cloud services as a real-time, on-demand supply and demand model.
In this model, the right sizing of resources can be more readily achieved because the fundamental premise is that cloud services should not be required to fulfill long-term requirements. They can scale their instances up and down based on demand — say, an enterprise’s sale peak periods, marketing campaigns, etc.
Strategically managing cloud costs is a tricky affair, and it’s extremely important to start off right. That said, the above-mentioned sources of potential overspending can be addressed at the outset. This not only helps slash costs but also simplifies the overall cloud migration process. Of course, continuous monitoring of the cloud ecosystem is a must here.
The article has been written by Anurag Sinha, Co-Founder & Managing Director, Wissen Technology