Edge traffic, always on your best path

Make the Internet choose your best path.

EROS continuously samples, analyzes, and safely optimizes your edge routing—so your traffic takes the fastest, most reliable, and cost-aware path.

Supported Vendors

CiscoJuniperAristaNokiaVyOSMikrotik

via BGP / NETCONF / gNMI

How EROS Works

EROS observes your live traffic and BGP state, probes each upstream, and safely biases routing toward the path that performs best—under your policies and cost caps.

Sample

Ingest NetFlow/IPFIX/sFlow + live BGP. Active probes per upstream. Target IP sets. Time-synced telemetry.

  • • Multi-provider source-based routing
  • • Paris-traceroute for path discovery
  • • Adaptive probe cadence
  • • Sub-ms timestamping (PTP optional)

Analyze

Per-prefix quality models. Right-sized prefix (/24→/21) selection. Anomaly detection. Policy guardrails.

  • • RTT / jitter / loss / stability scoring
  • • Path-equivalence grouping
  • • Dwell time & hysteresis filters
  • • Cost & commit awareness

Optimize

Scoped BGP injections (NO_EXPORT). LOCAL_PREF and provider communities. Automated rollback. Change budgets.

  • • Safe, scoped route preferences
  • • Provider-specific levers
  • • Automatic regression detection
  • • Cost cap enforcement

Example Policy (Pseudo-DSL)

policy optimize-latency {
  target: prefixes-with-tag("latency-sensitive")
  metric: rtt
  threshold: improvement >= 8ms for 5min
  max-changes: 2/hour/prefix
  scope: NO_EXPORT
  cost-cap: respect-95th-percentile
  rollback: on-regression or no-effect after 10min
}

The Appliance

Out-of-path by default; inline-capable when you want it. TPM-secured, hot-swap, PTP-optional timing.

EROS-100

1U Short-Depth

  • • Single site deployment
  • • Up to 5–10 Gbps traffic
  • • 2–3 upstreams
  • • 32 GB RAM
  • • 2× SFP+ ports
POPULAR

EROS-300

1U Full-Depth

  • • Medium site / PoP
  • • Up to 40 Gbps traffic
  • • 3–6 upstreams
  • • 64 GB RAM
  • • 4× SFP28 ports

EROS-700

2U Rackmount

  • • Large edge / data center
  • • Up to 100–200 Gbps traffic
  • • 6–12 upstreams
  • • 128 GB RAM
  • • 8× QSFP28 ports, HA clustering

Processing

Modern x86 (EPYC / Xeon-D) with AES-NI, AVX2

Storage

Mirrored NVMe for OS + timeseries DB

Networking

Dedicated OOB mgmt + data/probe ports

Security

TPM 2.0, Secure Boot, signed images

Integrations

Works with your existing infrastructure through standard protocols and APIs.

Router APIs

Control your edge routers through industry-standard protocols.

BGP (read + limited write)
NETCONF / gNMI / RESTCONF
SSH fallback for legacy devices

Provider Levers

Per-upstream community libraries for fine-grained control.

Prefer / de-prefer communities
AS-path prepending controls
RPKI validation (RTR client)

Telemetry Export

Send data to your existing monitoring and alerting systems.

Syslog / CEF to SIEM
Webhooks to Slack / Teams
REST / GraphQL API

Authentication

Enterprise-grade access control and audit logging.

SSO / SAML / OIDC
RBAC with approval workflows
Tamper-evident config bundles

Safety & Guardrails

Scoped BGP, conservative guardrails, automatic rollback, full auditability.

Change Hysteresis

Only shift routes when sustained improvement is detected.

  • • ≥ 8–10 ms sustained improvement
  • • ≥ 1.5× jitter improvement
  • • Sustained for ≥ 5 minutes
  • • Max 2 changes per prefix per hour

Blast Radius Limits

Prevent cascading failures with strict change budgets.

  • • Max N optimized prefixes per provider
  • • Most specific: /24 (tunable)
  • • Default NO_EXPORT on injected routes
  • • Never influence beyond local AS

Cost Awareness

Optimize performance while respecting your budget constraints.

  • • Respect 95th percentile / commit levels
  • • "Optimization under cap" mode
  • • Cost wall for expensive paths
  • • Override only for critical traffic

Automatic Rollback

Detect and revert changes that don't improve performance.

  • • Regression detection within 10 minutes
  • • No-effect detection and revert
  • • Full change audit log
  • • Before/after traceroute snapshots

High Availability

N+1 node cluster with leader election. Read-only mode on split-brain. Dual hot-swap PSUs. Automatic failover with state synchronization. Zero-downtime upgrades.

Use Cases

EROS delivers measurable improvements across diverse edge environments.

Multi-Homed Enterprises

Edge sites with 3+ upstreams running latency-sensitive SaaS, VoIP, or video conferencing. Automatically route traffic through the best-performing provider in real-time.

ISP & Colo Edges

Smooth out brown-outs on specific transit providers. Detect and route around congestion before customers notice. Maintain SLA compliance automatically.

Gaming & CDN Peering

Minimize latency for gaming traffic and CDN origin pulls. Optimize peering zone performance. Reduce jitter for real-time applications.

Multi-Site Federation

Deploy EROS across multiple sites with centralized policy management. Consistent optimization strategies with local autonomy and failover.

Results & Reporting

Measure the impact with comprehensive before/after analytics and provider scorecards.

RTT Improvement (ms)

Packet Loss (%)

-42%
Average RTT Reduction
-78%
Packet Loss Reduction
99.7%
Path Stability

Weekly Reports Include

• Provider scorecards with SLA compliance
• Per-prefix optimization deltas
• Path stability and flap analysis
• Cost vs. performance tradeoffs
• Root-cause summaries for incidents
• Change audit log with before/after traces

Pricing & Editions

Choose the edition that matches your operational requirements.

Essentials

Observe & Report

  • Full telemetry collection
  • Active probing & path analysis
  • Provider scorecards
  • Weekly reports
  • Web UI & API access
Contact Sales
RECOMMENDED

Pro

Auto-Optimize

  • Everything in Essentials
  • Automatic route optimization
  • Policy guardrails & rollback
  • Cost-aware optimization
  • Change audit logs
Contact Sales

Enterprise

Advanced & HA

  • Everything in Pro
  • High availability clustering
  • Multi-site federation
  • Cost model integration
  • 24×7 TAC support
Contact Sales

All editions include onboarding, vendor-specific golden configs, and software updates.

Talk to GridSite

Ready to operate on your best path? Get in touch to discuss your requirements or request a pilot.

Or reach us directly:

info@gridsitetech.example

Frequently Asked Questions