We share a few examples of real CloudZero customer anomalies, what caused them, and the action the customer took.
Please note, for customer privacy we have removed any identifying information.
One of the many aspects of CloudZero’s cloud cost intelligence is anomaly detection.
Unlike solutions which require you to set budgets or thresholds, CloudZero uses machine learning to automatically detect and alert on anomalies, then send them directly to dev team Slack channels where the responsible engineers can view and take action.
Here are just a few examples of real customer anomalies, what caused them, and the action the customer took.
Often when people think about cloud cost optimization, they often focus on their applications running in production. However, we often see runaway costs during R&D as well.
In this case, a data scientist was repeatedly running very expensive queries over a large data set — and they had no idea it was costing hundreds of dollars.
It triggered this alert in CloudZero, which was sent to Slack:
When they clicked “Review Anomaly” to investigate, they were able to quickly see the source of it in the CloudZero Cloud Cost Intelligence Platform — and understand when and why it happened.
After seeing the alert, the data scientist wrote a more efficient query and stored the results locally for later processing. Once the fix was made, the same data science project also delivered results faster allowing the company to deliver customer value sooner than originally planned.
In this example, the customer had accidentally misconfigured their VPC.
Every time they deployed code or pushed a new build, it was causing gigabytes of data that typically would have stayed within their environment to be pushed outside Amazon and back, causing much larger data transfer costs. The activity triggered this cost anomaly:
Each of those pink spikes is a deployment. The issue was fixed with a simple change to their terraform template in minutes.
In this example, a developer pushed a code change that caused files to be transferred instead of having their headers read, which caused data transfer costs to spike by over $1,300 in 11 hours.
Within hours, they received an anomaly alert. One of the engineers quickly investigated. Once it was fixed, they commented on the anomaly alert to let the team know it was fixed! This action saved thousands in ongoing cost.
If you want to ensure your engineering teams have visibility into the cost of their actions, check out CloudZero and request a demo today.
