ACTIVE DEVELOPMENT NOTICE: Website is currently in development — data might be inaccurate. GressTrace is 100% open source & self-hostable.

Product Capabilities

Built from the ground up for Kubernetes network egress observability, cost attribution, and incident response.

Blame-Trail Root Cause Engine

Correlates financial network spikes with Kubernetes deployment updates, container image tags (`cart-service:v2.1.0`), HPA scaling actions, and Git commit SHAs within a ±30 minute temporal window of any anomaly.

NAT Gateway & Cross-AZ Cost Isolation

AWS charges $0.045/GB for NAT Gateway processing fees and $0.02/GB for cross-AZ data replication. GressTrace attributes out-of-band NAT charges down to specific pod destinations and namespaces.

Passive Zero-Overhead Telemetry

Python and Go client SDKs feature a non-blocking in-memory ring buffer, background thread flushing, Gzip payload compression, and automatic exponential backoff retries with zero impact on core application execution paths.

Statistical Noise-Floor Suppression

Dual-pass Z-Score ($Z = \frac{X - \mu}{\sigma}$) and percentage change ($\Delta\%$) statistical detection algorithms ignore routine baseline fluctuations and alert Slack, Webhooks, or Email when genuine cost spikes trigger.

Multi-Cluster Data Scoping

Enables engineering teams to isolate, group, and query network egress costs across AWS EKS, GCP GKE, Azure AKS, and bare-metal Kubernetes clusters with zero data cross-contamination.

Automated Telemetry Retention Cleanup

Daily background Celery tasks automatically purge telemetry snapshots older than configured project retention limits (default 30 days), preventing database storage bloat in self-hosted deployments.

The product is currently in active development phase. Self-hosting instructions and preview source code are available on GitHub.