Course contentsShow
← Security
Lesson 1915 of 3,05238. Cloud Security - Monitoring and IRFree lesson

Evidence Identification and Preservation in Cloud

Learn techniques for identifying and preserving volatile cloud evidence before resources are terminated or logs are rotated.

Evidence Identification and Preservation in Cloud

What you'll learn: How to quickly identify and preserve volatile cloud evidence before it disappears due to auto-scaling, resource termination, or log rotation.

The Challenge of Cloud Evidence

In traditional data centers, you could physically secure a server and preserve its hard drive. In the cloud, resources are ephemeral—instances auto-scale down, containers terminate after crashes, and logs rotate automatically. Evidence literally evaporates if you don't act fast.

Think of cloud evidence like footprints in sand near the tide: you must capture them before the waves (automation) wash them away.

Critical Volatile Evidence Sources

Memory and Instance State

When an EC2 instance or VM is compromised, its memory contains running processes, decryption keys, and active network connections. Once terminated, this data is gone forever. You must snapshot the instance or capture memory dumps before containment actions destroy it.

Ephemeral Logs

Cloud services often rotate logs every few hours. VPC Flow Logs, application logs, and container stdout/stderr might only retain 24-48 hours of data unless explicitly configured otherwise. Identify your retention windows immediately during an incident.

Temporary Resources

Serverless functions, autoscaling groups, and spot instances come and go constantly. The compromised Lambda function might already be replaced by a new version. Preserve function code, environment variables, and execution logs before they're overwritten.

Preservation Techniques

Snapshot everything first: Create disk snapshots, export logs to immutable storage, and tag resources with "DO NOT DELETE" policies before analysis.

Use APIs for automation: Leverage CloudTrail, GuardDuty findings, and flow logs to programmatically identify suspicious resources the moment alerts fire, then trigger automated preservation workflows.

Isolate, don't terminate: Instead of deleting a compromised instance, change security groups to deny all traffic while preserving its state for forensics.

Key Takeaway: Cloud evidence is volatile by design—identify critical data sources immediately, preserve them through snapshots and log exports, and isolate rather than destroy compromised resources to maintain forensic value.