Best Practices

Agent Deployment Strategy

Production Servers

  • Managed agents in Health Only mode by default (<1% CPU, ~50 MB RAM)
  • On-Demand for maintenance windows; Full Monitoring only during active incidents
  • Standard (Baseline) provider configuration; alert rules for CPU, memory, disk

Dev/Test Systems

  • Managed for AI diagnostics, desktop for basic visibility
  • Comprehensive config for debugging sessions

User Workstations

  • Desktop agents (free) fleet-wide; managed only for VIP/critical workstations
  • On-Demand collection only when troubleshooting

Cost Optimization

  • Desktop agents for non-critical systems (free, unlimited)
  • Health Only mode when not troubleshooting
  • Right-size subscription tier to actual query usage
  • Consolidate agents in single org for volume discounts
  • 15% savings with annual billing
  • BYOK tier for heavy AI users
  • Delete inactive agents promptly

Security

  • Use registration tokens for all agent installations; never share tokens in plaintext outside secure channels
  • Create separate tokens per environment (production, staging, dev) and per deployment wave
  • Revoke tokens immediately when a deployment is complete or a token is compromised
  • Set max-agent limits and expiry dates on tokens to limit blast radius
  • Separate orgs for production vs. non-production
  • Review team access regularly; remove departed employees and cancel their admin-purchased subscriptions
  • Exclude sensitive paths from ETW collection
  • Route critical alerts to security team channels

Alert Tips

  • Use duration requirements to avoid false positives on brief spikes
  • Test notification channels immediately after creation
  • Multiple channels for critical alerts (email + Slack + webhook)
  • Tag agents to target rules at specific groups
  • Review rules monthly based on alert frequency

Getting Help

Response Times: Professional 24hr | Business 8hr | Enterprise 2hr

Roadmap

Near-Term (3 months)

  • Enhanced alert templates, alert analytics, agent group management, multi-agent timeline UI

Mid-Term (3–6 months)

  • Mobile app with push notifications, extended retention, custom dashboards, AD auto-discovery, integration marketplace

Long-Term (6–12 months)

  • ML anomaly detection, predictive alerting, automated remediation, Linux/macOS agents, Kubernetes monitoring, public API

Timelines subject to change based on customer feedback.