Moving to the cloud can save businesses money, but the reality is that without a good cost optimisation strategy, overspending can occur. Stacklet’s ‘State of Cloud Usage Optimisation 2024’ shows that companies are losing billions in cloud waste every year. More than three-quarters (78%) of the respondents estimated that up to 50% of their cloud spend is underutilised.
Cloud is easy to mismanage, but there are multiple solutions to get costs under control. All the hyperscalers have free native cloud management tools. AWS has Cost Explorer, which provides detailed insights into AWS spending. There is also Azure Cost Management, which helps optimise Azure cloud costs, and Google Cloud’s Cost Management Tools, which offer tools for tracking cloud expenses.
There are also any number of FinOps tools to help organisations manage cloud costs. And yet, with all these tools, why are businesses still overblowing their cloud budgets? “The most effective tools depend on what you’re trying to optimise,” says Dirk Ras, architect in the office of the CTO, practice lead: cloud, at Dariel Software. “If the wrong technology is used for the wrong purpose, costs can escalate fast.”
Fragmented ecosystem
According to Flexera’s ‘2024 State of the Cloud Report’, around 89% of organisations are now using a multicloud approach. Native tools work, but it gets tricky when you need to manage more than one. The benefits of using multiple private and public cloud platforms are a given – multicloud means avoiding things like vendor lock-in and being able to use the right cloud for the right application. But there are some harsh realities with multicloud that only crop up when monthly bills start to come in, particularly when the hyperscalers are releasing a constant stream of tiers and tools at different price points.
Cloud optimisation often falls into the cracks between IT, finance and application owners, leading to a lack of clear accountability.
Alex Russell, Nutanix
“For cloud optimisation to be effective, the most important thing is knowing what you need and picking the right technologies from the beginning,” says Ras. “It makes a big difference.” All the cloud providers have cost calculators, and the more detail you put into them, the more accurate the estimates will be. If it’s an optimisation exercise for an existing cloud environment, Ras says that the first step should be an audit. “You need to check whether the allocated resources match actual usage,” he says. “There’s no point in running a massive instance for a web server that only gets a couple of hits a day.” Once that’s done, optimisation points can be tackled, such as auto-scaling, instance right-sizing or adjusting storage.
Nasty surprise
FinOps should be a shared responsibility between the cloud team, finance and operations, says Ras, “because costs need to be managed properly across all three.” Strong governance, proper tagging and financial tracking make it easier to reconcile costs. In large businesses, different teams deploy their own resources, but without centralised governance, this can result in underutilised or redundant services. A good example is when teams spin up resources without realising how much they cost, and finance doesn’t have real-time visibility into cloud spend. You cannot control what you cannot see.
“That’s why businesses only see the real impact at the end of the month, sometimes with a nasty surprise,” says Ras. This is why Alex Russell, regional sales manager, SADC, Nutanix, recommends making use of a cost governance solution that can deliver detailed insights into cloud spending and resource utilisation. “Cloud optimisation often falls into the cracks between IT, finance and application owners, leading to a lack of clear accountability,” he says. Cost governance and compliance tools help ensure that an organisation’s cloud spend is in line with its policies and regulations. He says businesses will need to use a combination of tools in their optimisation efforts.
“Without clear cost allocation, it’s hard to hold teams accountable,” says Vikar Singh, a cloud solutions architect at Cloud Essentials. “Lack of governance leads to uncontrolled resource growth, meaning you pay for services you don’t need.” He adds that accurate data is important to identify savings opportunities.
With reliable data, businesses can also begin to harness AI, which is now something all the popular cloud cost optimisation tools like Cast AI, CloudZero and Harness are looking at.
Picking the right technologies from the beginning makes a big difference.
Dirk Ras, Dariel Software
AI tools can look at a business’ past data, see how cloud services have been used and spot trends to predict future expenses. When a cloud environment is being continuously analysed, organisations can plan their budgets better and steer clear of surprise costs.
With overprovisioning being a common budget disruptor, it will help to focus on usage. A business could have underutilised or idle resources, such as unused virtual machines, storage volumes or over-provisioned compute instances. “The complexity of the design of the cloud platform makes it harder for people without sufficient skills or experience to identify areas for optimisation,” says Richard Vester, managing executive: EMEA AWS practice at iOCO.
AI helps to pinpoint potential issues and make recommendations for elimination. IBM Turbonic, for example, creates a digital twin of a firm’s entire infrastructure, mapping cloud resources in real-time. This digital twin acts like a virtual marketplace where applications bid for the resources they need. The AI engine tracks actual usage against provisioned capacity and quickly spots overprovisioned VMs and containers. This makes it easier to proactively implement cost-saving strategies while balancing (or resizing) workloads across instances. With AI in the mix, cloud cost optimisation can be less of a shock, which is exactly what you want when handling cloud finances.
That said, it’s worth keeping in mind that tools enable people, but don’t replace them. “An organisation needs someone to actually log in and review dashboards, interpret the optimisation recommendations, drive the necessary changes across teams, and own the ongoing cost management programme,” says Verster. “A combination of the right tools and the right expertise is the only way to improve the performance and economics of the cloud.”
WHO ARE THE BIGGEST CLOUD OFFENDERS?
One of the primary reasons cloud services go to waste is over-provisioning. A lot of businesses still approach cloud the same way they did on-premises, provisioning more than they need “just in case”.
“But the whole point of cloud is elasticity – it’s easier to scale up than to scale down,” says Dirk Ras, architect in the office of the CTO, practice lead: cloud, at Dariel Software. “People forget just what utility-based computing actually means,” says Richard Vester, managing executive: EMEA AWS practice at iOCO. No one will leave a tap running when they have finished washing their hands, and yet organisations leave many of their cloud services running forever – even if they aren’t being used. Verster says companies should switch unused services off, or put them on schedules when they need to be available intermittently. Another big issue is orphaned resources. When projects end, the resources attached to them often don’t get cleaned up, so things like databases and instances keep running when they’re no longer needed. A lack of auto-scaling and right-sizing can also lead to cost issues.
Cloud platforms all have built-in auto-scaling, but many organisations don’t use it properly. Storage mismanagement is another factor. “Why store large amounts of data in a database when something like S3 would do the job at a fraction of the cost?” asks Ras. Then there’s licensing. Because some cloud services come with built-in licensing and others don’t, businesses sometimes end up paying for both without realising it. Shadow IT is not a new issue, but in cloud, its risks and impact have magnified significantly. When teams bypass governance and spin up their own cloud resources, it can lead to both cost and security risks. “A lot of people don’t think about it this way, but cloud cost anomalies can also be an early indicator of security risks,” says Ras. “If your costs suddenly spike without a good reason, that’s usually a sign something needs investigating.”
* Article first published on brainstorm.itweb.co.za
Share