Three scenarios
| Persona | Challenge | How Fleet Management helps |
|---|---|---|
| Platform team | 200 collectors, multiple environments, coordinating changes is slow | Create once, deploy everywhere with attribute matching |
| Growing startup SRE | Went from 5 to 40 collectors, manual management doesn’t scale | Enterprise-grade control without building custom tooling |
| DevOps during incident | Need profiles now, can’t wait for change management | Create temporary pipeline, toggle on/off instantly |
Common patterns
- Centralized control — One team manages fleet for entire org
- Environment segmentation — Different configs for prod/staging/dev
- Gradual rollouts — Test new configs on subset before fleet-wide rollout
- Temporary debugging — Collect targeted telemetry during incidents, remove after
Script
This step explains who uses Fleet Management, how they use it, and where you might fit.
First, meet the platform team. They’re responsible for observability infrastructure across the entire organization. They have 200 collectors spread across production, staging, and development environments. Different teams own different services, but the platform team owns the collectors. Before Fleet Management, rolling out a new metrics scrape meant coordinating with every team, updating Helm charts or Ansible playbooks, and hoping nothing broke. Now they create one pipeline, set the matchers, and it rolls out automatically. They’ve gone from week-long rollouts to same-day changes.
Next, consider the SRE at a growing startup. They started with 5 collectors and managed them manually. Now they have 40, and the manual approach doesn’t scale. They don’t have time to build custom tooling. Fleet Management gives them enterprise-grade fleet control without the engineering investment. They can focus on reliability instead of configuration management.
Finally, think about the DevOps engineer responding to an incident. Production is having issues, and they need to start collecting profiles immediately. They can’t wait for a change management process. With Fleet Management, they create a profiling pipeline, match it to the affected collectors, gather the data they need, then deactivate it. No deployments, no rollbacks, no cleanup—just toggle it off when done.
These are the patterns we see. Central control, scaling without custom tooling, and rapid response to changing needs.