Skip to main content
Tracer/sweep is a fast, systemwide scan that uncovers cloud compute waste using observed execution behavior, not estimates, tags, or predictive shutdown heuristics. While cloud billing tools report what exists and what is billed, Tracer/sweep looks at what actually ran. It surfaces inefficiencies that do not appear in Cost Explorer, dashboards, or manual reviews, without stopping instances or changing workloads.

​
What Tracer/sweep does

Tracer/sweep analyzes execution activity across your cloud compute environment to identify waste that is otherwise hard to detect. It:
  • Scans compute based on real runtime activity
  • Identifies inefficiencies that persist across runs and instances
  • Surfaces opportunities for cost reduction with minimal investigation
  • Delivers results quickly, without requiring long tuning cycles
Tracer/sweep is designed for fast time-to-value, especially when cloud spend is high but existing reports cannot explain why.
Idle resource detection showing unutilized compute

​
What makes Tracer/sweep different

Most cloud cost tools rely on:
  • Billing aggregates
  • Tags and allocation rules
  • Coarse utilization metrics
  • Predictive or AI-driven shutdown recommendations
These approaches often miss execution-level inefficiencies or introduce risk by acting on incomplete information.
Tracer/sweep is different because it:

Execution-based

Relies on observed execution and activity patterns

Deep visibility

Detects waste inside instances that appear “active”

Safe

Avoids predictive shutdown heuristics that can cause errors

Zero setup

Requires no tagging discipline or manual classification
Tracer/sweep bases its findings on what the system actually did, not on what it was configured to do.

​
What Tracer/sweep finds

Tracer/sweep identifies several common classes of hidden waste:
Instances that are running and billed, but spend significant time doing no useful work.
Tracer/sweep detects idle resources that remain active and billed, including:
  • EC2 instances that are running but not executing meaningful work
  • AWS Batch workers that are provisioned but stuck idle
  • Instances with negligible CPU activity over extended periods
These resources often appear “active” in cloud dashboards but contribute no useful execution. By relying on observed execution activity rather than instance state alone, Tracer/sweep highlights resources that can be safely reviewed for shutdown or resizing.
Compute that appears active due to long runtimes or background activity, but shows low real utilization.
Instances or workers that remain active without meaningful execution over time (where observable).
All findings are derived from execution signals, not billing metadata alone.

​
Outputs

Tracer/sweep produces:
  • Ranked cost-saving opportunities, prioritized by impact
  • Suggested actions, such as resizing, consolidating, or retiring resources
  • Estimated impact ranges, based on observed usage patterns (when available)
Results are explainable and grounded in execution behavior, making them easy to validate before acting.

​
What Tracer/sweep does not do

Tracer/sweep is intentionally conservative.
It does not:
  • Automatically stop or terminate instances
  • Replace FinOps platforms or cloud billing tools
  • Require perfect tagging or cost allocation hygiene
  • Make decisions based on prediction alone
Tracer/sweep surfaces evidence. Teams remain in control of what actions to take.

​
Requirements

Tracer/sweep requires Tracer/collect to be installed so execution activity can be observed. It operates using:
  • Execution signals captured by Tracer/collect (Not required, but recommended)
  • Cloud cost and usage data from supported providers (AWS currently supported)

​
Install Tracer/sweep

Connect your AWS account in the Tracer UI and get results in minutes. Tracer/sweep requires read-only access to your AWS account to analyze cloud usage and execution activity. Setup takes a few minutes and does not modify any running resources.

​
Prerequisites

  • An AWS account with permission to create IAM roles
  • Access to the Tracer UI
  • No agents, code changes, or tagging required
1

Create a read-only IAM role

Tracer/sweep uses a dedicated, read-only IAM role to access cost and execution metadata.
  1. Open the AWS CloudFormation link provided in the Tracer UI.
  2. Deploy the CloudFormation stack in your AWS account.
  3. The External ID is pre-filled and does not need to be changed.
This stack creates a read-only IAM role scoped to the permissions Tracer/sweep requires.
2

Enter the role ARN

After the CloudFormation stack finishes deploying:
  1. Copy the Role ARN from the CloudFormation output named TracerReadRoleArn.
  2. Paste the role ARN into the Role ARN field in the Tracer UI.
  3. Save the configuration.
Tracer will access your account using this role.
3

Review results

Once connected, Tracer/sweep immediately begins to analyze your environment. You can:
  • Review up to 90 days of cloud spend
  • See what is running now
  • Identify idle or underutilized resources
  • Estimate potential cloud waste based on observed execution activity
Results appear directly in the Tracer UI and update automatically.

​
Security model

  • The IAM role is read-only
  • Tracer does not start, stop, or modify resources
  • No application data, environment variables, or secrets are accessed
  • You can revoke access at any time by deleting the IAM role

​
What happens next

After setup:
  • Tracer/sweep highlights ranked cost-saving opportunities
  • Findings are based on observed execution behavior, not predictions
  • You remain in full control of any actions taken
To optimize specific pipelines and workloads, continue with Tracer/tune.

​
Summary

Tracer/sweep provides a fast, reliable way to uncover cloud waste using real execution behavior. By focusing on what actually ran, rather than what was tagged or predicted, it reveals inefficiencies that other tools miss and does so without introducing operational risk.

Tracer/collect
Learn how execution signals are captured

Tracer/tune
Optimize specific pipelines and workloads