5 results found
-
Supporting Threads functionality for DoiT Anomaly detection
Allow DoiT Anomaly detection users to create and manage threads for the the actions required to triage and resolve anomalies.
4 votes -
Enhance daily anomaly notification by adding granularity about the usage the timing when the spike actually happened
Add the context in the anomaly notification which is like highlighting the usage and cost impact along with the timing when it occurred during the day
An example: Hey, between 2-3AM on May 5th you had 150M requests which lead to a charge of 110$ and this is why we triggered the anomaly"
2 votesWe already support natural language summarization via Ava
-
Anomaly detection on the Usage data
Expand our existing cost anomaly detection using the billing data to also support the anomaly detection on the usage data
1 voteWe will reconsider the priority of this feature later in this year, hence declining the request for now.
-
Cost Anomaly Detection - Exclude Attribution from algorithm
Hello Team,
Customers currently can add an Attribution for Anomaly Detection to monitor abnormal spend behavior on a specific segment of their business. A customer just surfaced the need of excluding a specific Attribution from the Anomaly Detection logic due to the nature of that segment of the business, where they are expecting those anomalies to happen. The desired outcome would be that all infrastructure linked to that Attribution would not be flowing into the Cost Anomaly algorithm, hence the anomalies would not get surfaced as per customer preference.
Do you think this would be possible?
Kind Regards,
1 voteCan be done currently by disabling default anomalies and monitoring for anomalies an attribution that excludes the desired selection.
-
Anomaly detection by custom cost tag
We'd like to adjust our anomaly detection to allow selection of custom cost tag.
For instance, our primary cost tag is the Component tag, we now often get SKU-based anomalies that aren’t anomalies when looking at en entire Component tag value.
For instance: We may have a cluster tagged with Component=EMRCluster1 that switched from using one type of EC2 instance to the other. This will show up as an anomaly, but the entire service under this Component tag didn't necessarily have an anomaly.
30-50% of the anomalies we get now are now red herrings because of this.
My questions:
-…1 voteThis is currently possible using the allocations. We understand that manual work is required to create multiple allocations however it is just a one time effort. We are declining it for now as it doesn't take higher priority than the other capabilities planned for anomaly detections however we will be revisiting this in the future once we get more customer data.
- Don't see your idea?