This post explains how to design and implement a compliance-ready logging architecture to satisfy NIST SP 800-171 Rev.2 / CMMC 2.0 Level 2 control AU.L2-3.3.2 in the context of the Compliance Framework, focusing on practical, actionable steps for small businesses that process Controlled Unclassified Information (CUI).
Key objectives of AU.L2-3.3.2 (Compliance Framework perspective)
The core objective of AU.L2-3.3.2 is to ensure that systems generate and retain sufficient, reliable audit records so you can detect unauthorized activity, support forensic investigations, and demonstrate compliance with the Compliance Framework. That means logging the right events, protecting log integrity and availability, centralizing storage, enforcing retention and access controls, and establishing review processes so logs are actionable.
Designing a compliance-ready logging architecture
A practical architecture consists of (1) log sources (endpoints, servers, network devices, cloud services, applications), (2) local collectors/agents (Winlogbeat, NXLog, rsyslog, Fluentd), (3) secure transport (TLS syslog, HTTPS), (4) a centralized log store/SIEM (Elastic Stack, Splunk, Sumo Logic, cloud-native), (5) immutable/archival storage (WORM S3/Blob with object lock or write-once backup), and (6) analytics/alerting and review workflows. For Compliance Framework mapping, document each component and the control(s) it supports: collection, protection, retention, review.
Log sources and collection — what to capture
Capture authentication and account management events (successful/failed logins, privileged use), system changes (config, service start/stop), application exceptions, data access to CUI repositories, network gateway events (VPN, firewall), endpoint EDR alerts, and cloud provider control-plane logs (AWS CloudTrail, Azure Activity Log). Example: a small business with an on-prem Windows domain and AWS-hosted applications should collect Windows Security and Sysmon events, Active Directory logs, CloudTrail, VPC Flow Logs, firewall syslog, and application logs from containers or web servers.
Transport, normalization, and secure storage
Use encrypted transport (Syslog over TLS RFC 5425, Winlogbeat/Beats over TLS, HTTPS/Fluentd) to move logs to your collector. Normalize logs into a common schema (ECS, CEF) to make alerting consistent. Store active logs in a SIEM for search and alerting, and archive to an immutable store (S3 with Object Lock + SSE-KMS or Azure Blob immutable storage). Enforce server-side and key management policies: use KMS keys with strict IAM, separate key admin and log admin roles, and enable S3 bucket policies to block public access and enforce TLS-only PUT.
Retention, integrity protection, and access control
Define retention aligned to contract and incident response needs — common practical baselines are 90 days in hot storage for daily detection, 180–365 days in warm storage for investigations, and multi-year cold archives if contracts require. Protect integrity by hashing log files (SHA-256) and storing hash manifests in a separate, write-only location or an HSM-backed ledger. Implement role separation: only a few admins can manage log configuration; read access is granted to IR/forensics teams via a documented request process; enforce MFA and just-in-time elevation for log admin access.
Alerting, monitoring, and review processes
Create SIEM detection rules for common indicators: repeated failed logins, privilege escalation, unexpected service account activity, large data exfil attempts, or anomalous API calls in cloud logs. Define a daily/weekly review cadence for high-risk alerts and a monthly audit of logging configuration and coverage. For small teams, build playbooks that map common alerts to triage steps (source verification, snapshot preservation, escalation) and automate evidence collection (log export, kernel memory snapshot templates) to speed investigations while preserving chain-of-custody.
Implementation steps and a small-business scenario
Step-by-step for a small business (example: Acme Consulting with Windows domain + AWS): 1) Inventory systems that touch CUI. 2) Enable Windows Event Forwarding or deploy Winlogbeat to forward Security and Sysmon channels to a hardened collector. Example Winlogbeat config snippet: event_logs: - name: Security - name: Microsoft-Windows-Sysmon/Operational. 3) Enable AWS CloudTrail for all regions and send logs to an encrypted S3 bucket with Object Lock: aws cloudtrail create-trail --name AcmeTrail --s3-bucket-name acme-cloudtrail-logs. 4) Configure Syslog-ng/rsyslog with TLS to collect firewall and network device logs. 5) Route logs to an Elastic Stack or cloud SIEM, normalize fields, and enable retention lifecycle policies. 6) Archive logs to S3 Glacier or Azure Archive with immutable settings. 7) Implement hashing and periodically verify signatures. 8) Document and test restore/forensic playbooks quarterly.
Risks of not implementing AU.L2-3.3.2 and compliance tips
Without a compliant logging architecture you risk delayed breach detection, inability to investigate and attribute incidents, loss of CUI, audit failures, contract termination, and financial or reputational damage. Tips: map each control requirement to a specific architecture element and evidence artifact (e.g., CloudTrail = evidence of API logging), perform gap assessments, automate evidence collection for audits, and run tabletop exercises to validate that logs and playbooks produce usable evidence. Keep a changelog of logging configuration changes as evidence that you monitor the logging pipeline itself.
Summary: Build a layered logging architecture—identify sources, secure and normalize transport, centrally store with immutable archives, enforce retention and access controls, and operationalize detection and review—so you can demonstrate to auditors and stakeholders that your environment meets the Compliance Framework requirements of AU.L2-3.3.2. Start small (capture the high-value log sources), automate where possible, document everything, and iterate until you have reliable, reviewable audit records that support detection and investigations.
