AWS has launched a new machine learning feature in its Cost Management suite to help customers mitigate nasty surprises on their cloud bills.
Now in preview, AWS Cost Anomaly Detection uses machine learning to understand a customer's spending patterns and send alerts when it finds anomalies, such as a large one-time jump or a persistent rise in costs over time, according to AWS.
The tool is embedded within the main AWS Cost Management dashboard and spots anomalies according to AWS services; individual accounts or groups of linked accounts; cost centers, such as engineering; and cost categories, where spending is mapped to a company's organizational structure, AWS said.
The system sends anomaly reports to customers via email or AWS SNS (Simple Notification Service). Customers can throttle the number they receive based on estimated cost thresholds. Anomaly Detection provides root-cause analysis for anomalies and is integrated with AWS Cost Explorer, which provides graphical views.
Anomaly Detection underscores native vs. third-party cost tool question
AWS Cost Management's other features include tools for budgeting, cost and usage reports, and cost categorization. It is priced according to usage, with a cost of $0.01 per emailed budget report or request to the Cost Explorer API. Planned pricing for Anomaly Detection isn't publicly available.
Anomaly Detection's use of machine learning relieves some of the legwork around cost containment from customers and moves AWS into an area served by many partners, such as CloudZero, a company based in Boston. CloudZero doesn't necessarily view the new AWS tool as a competitive threat, however.
"CloudZero's strategy has always been to embrace and extend the services AWS provides to deliver unique value for SaaS and digitally-native businesses building on AWS," CloudZero founder and CTO Erik Peterson said in an email, citing CloudZero's integration with AWS's CloudWatch observability and monitoring service.
Jean AtelsekAnalyst, 451 Research
In addition, CloudZero's anomaly detection capabilities go beyond AWS' "entry-level" offering, he said. These features include an integration with Slack to deliver alerts and automatic categorization of cloud spending on features, products and teams, according to Peterson.
Other vendors offering cost anomaly detection for AWS include Cloudwiry and Anodot.
Overall, the choice AWS customers have for cost management tools on the platform comes with tradeoffs, said Jean Atelsek, an analyst with 451 Research.
"There are plenty of third-party alternatives, but none of them has the access to customer usage and billing data on its platform that AWS does," she said. "Perhaps the company can use that to its benefit for customers that are 'all-in on AWS' and plan to stay that way."
However, "the advantage of third-party tools is that they are not bound to AWS," Atelsek added. "If the customer gives access to their billing APIs on multiple clouds and on premises, they can set up thresholds in one place to get a cost view across the entire estate."