Skip to main content

Performance & Scale

Breeze Agent minimizes resource and network impact, supports large-scale rollouts, and can be tuned for different host performance requirements. Use this guide to plan and operate Breeze Agent deployments across large host estate.

Runtime and Resource Usage​

  • Typical runtime per cycle: < 60s for standard discovery workloads.
  • CPU and memory footprint: low; see Requirements for baseline sizing.
  • Scheduling:
    • Default cadence is ~15 minutes with jitter to avoid synchronized spikes.
    • Agents sleep between runs; there is no long‑lived daemon process by default.

Network Impact​

  • Breeze sends compact payloads of OS facts; typical daily volume is only a few MB per host at default cadence.
  • Where possible, plugin logic minimizes redundant data to avoid unnecessary bandwidth usage.
  • For environments with strict egress controls:
    • Ensure outbound TCP 443 is allowed to Breeze endpoints.
    • Consider placing Breeze behind proxies or TunHub for private networks.

Rolling Out at Scale​

When deploying Breeze to hundreds or thousands of hosts:

  • Use rings or waves:
    • Start with a pilot (for example, 5–10% of the estate).
    • Monitor health and Breeze: Last Update metrics in Cloudaware before expanding.
  • Leverage device management tooling:
    • AWS Systems Manager, Intune, MECM (ex-SCCM), and similar tools are covered in Fleet Deployment.
    • Use maintenance windows and rate limits where available.
  • Monitor capacity:
    • Track Breeze‑related CPU/memory usage on representative hosts.
    • Watch for scheduler backlogs (missed or overlapping runs) on heavily loaded systems.

Tuning Recommendations​

  • Prefer the default 15‑minute cadence unless there are strict constraints.
  • For very latency‑sensitive systems:
    • Run Breeze less frequently (for example, every 30–60 minutes) by adjusting the timer/cron schedule.
    • Avoid running heavy plugins during peak business hours.
  • For troubleshooting or short‑term investigations:
    • Temporarily increase logging verbosity, then revert to normal levels once resolved.