Asim Razzaq is CEO of Yotascale. Views are the author's own.
A third of senior IT leaders in a survey last year said they spent 20-40% more on cloud costs than they budgeted and two-thirds said “cost management and containment” is their main worry when it comes to running big data cloud technologies and applications.
For CFOs, getting a handle on cloud spend is core to their ability to budget and manage payables effectively. Given the risk of accidental spikes in cloud costs and misconfiguration issues, CFOs can face a huge cloud bill at the end of the month, especially as cloud-native services and on-demand pricing models take root.
Fortunately, thanks to their cloud access and relationship with the provider, and their team’s role on the contracts and payments side of cloud services, CFOs have a unique perspective on, and ability to manage, cloud costs. But that starts with an understanding how misconfigurations can throw off cloud spend.
Spinning up the wrong number of services
Misconfigurations are a mismatch between your cloud needs and how you’re actually using it. The development team could have provisioned a large instance for application testing that is no longer used and left idle, for example, or they're trying out newer cloud-native app development technologies. Or you could have frozen serverless calls, or maybe the development team just isn’t using reserved instances efficiently.
All of these simple misconfigurations can spin up the wrong number of services and instances, causing cost anomalies in the bill. If this spending appears small, it may go unnoticed for a long time. But the costs add up.
While cloud-based resources are great for scaling an organization, misconfiguration, unused resources, malicious activity and overambitious projects can cause spikes in resource use and costs. An obvious solution is just hiring more IT staff, cloud architects, a data scientist, and a smart engineer to fix it. While this expensive dream team might be able to reduce human error, if there is any malicious activity, this same team will then only be able to hand you a shockingly high cloud bill.
The solution is early detection and anomaly correction.
Enterprises can protect themselves from spending anomalies by continuously monitoring and optimizing cloud costs. A platform that can do this must be able to track cloud expenditure, detect anomalies, analyze their root cause, and alert potential issues. The platform should then give visibility into what went wrong to prevent similar future mistakes. It should identify these anomalies in real-time, accurately, and while maintaining a balanced budget.
Implementing continuous cost monitoring and anomaly detection is vital for dynamic cloud environments. Cloud cost anomaly detection works by analyzing cloud expenditure trends and forecasting spending behavior. It’s essential to identify any activity that doesn’t align with anticipated expenditure or deviates from the established pattern. With proper monitoring, organizations can get insights into what resources might be causing the deviation and take corrective measures before it’s too late. For example, you can shut down resources that are no longer in use or optimize resources that are costing too much.
Amazon Web Services offers a variety of tools to help their customers understand and manage their cloud costs, namely AWS Cost Explorer to understand cloud billing, and AWS Cost Anomaly Detection which enables users to detect, assess and evaluate unexpected cost anomalies in their AWS cloud services.
While these tools are helpful for small, simple cloud deployments, they aren’t sufficient to manage large scale deployments with multiple accounts and modern cloud architectures, including containers and multi-cloud deployments.
AWS Cost Explorer focuses on AWS billing, but it doesn’t provide any information in the context of a business’s organizational structure, and only updates once every 24 hours. Also, like every other analytics package, AWS Cost Explorer is dependent on the data it receives. Cloud cost attribution granularity depends on how you’ve assigned features such as namespace, labels, and tags. Without making the appropriate hierarchies by setting and maintaining tags, it’s challenging to analyze costs at a meaningful granularity level, and the effort and time required hinder efforts to achieve cost savings.
AWS Cost Anomaly Detection integrates with AWS Cost Explorer, leveraging machine learning to detect changes in cloud spend, analyze their root cause, and send alerts to these potential issues. However, enterprises may still need to dig through logs to discover what happened and fix the problem, and without correct tagging that is linked to the business’ organizational structure, the tool won’t be able to match the spend to the correct team and notify the people who can take action.
Cost optimization isn't a manual fix. Cloud cost management and optimization tools can help save enterprises millions of dollars a year in cloud spend. Tackling wasteful cloud usage and exploding cloud spend is the first step of cloud management. Major cloud users will need a dedicated individual to manage cloud costs and inefficiencies to get the most value out of their public cloud investment.