Skip to Content
Using AlertDAnalyzing Results

Analyzing Results

AlertD returns two complementary views for every query: structured table data and AI-generated insights. The table shows raw data (resource IDs, metric values, configurations) that you can sort, filter, and export. The AI insights provide context, identify patterns, and offer recommendations based on that data. Start by reading the AI summary to understand what AlertD found, then dive into the table when you need specific details, exact values, or want to verify the AI’s conclusions. Both views are essential—tables give you precision, insights give you direction.

When analyzing performance metrics, pay attention to CloudWatch statistics (Average, Maximum, Minimum, Sum) and time ranges. An EC2 instance at 100% CPU for five minutes is very different from sustained 100% over 24 hours. For cost optimization queries (unattached volumes, underutilized instances), prioritize by impact—focus on the largest, oldest, or most expensive resources first. For security findings (public buckets, unencrypted volumes), assess risk by data sensitivity and compliance requirements. Always combine AlertD’s analysis with your knowledge of the infrastructure: a “low CPU” instance might be correctly sized for burst workloads, or a “public” S3 bucket might be intentionally serving a website.

Use follow-up questions to build a complete picture. If AlertD identifies underutilized instances, ask about their tags, Auto Scaling group membership. Cross-reference results across multiple queries to avoid mistakes—for example, verify that an “idle” instance isn’t actually handling important batch jobs by checking disk I/O or network traffic. When results don’t make sense (empty results, unexpected values, contradictory recommendations), check the query scope (region filters, time ranges) and inspect the execution plan’s data streams to understand what AlertD actually queried.

Best Practice: Combine AlertD’s insights with your domain knowledge. The AI provides data analysis and pattern recognition, but you understand your business context, deployment schedules, and operational requirements. Trust the data, verify the recommendations, and use your judgment before making infrastructure changes.

Next Steps

Last updated on
AWS QueriesAWS