Frequently asked
questions

When your application issues a fill request, it reaches the nearest NetfillX edge node (via anycast). Our routing engine then evaluates the request in this order:

1. Rule matching — Static rules you define (geo-based, header-based, weight-based).
2. ML optimization — The Route Optimizer adjusts weights dynamically based on real-time endpoint health, latency, and historical patterns.
3. Circuit-breaker — Endpoints with elevated error rates are temporarily excluded; traffic redistributes automatically.
4. Dispatch — The selected route is returned to your client (or, with proxied mode, the request is forwarded directly by NetfillX).

The entire process — from receipt to routing decision — takes under 800 microseconds at p50 globally.

NetfillX supports these protocol families out of the box:

• HTTP/1.1, HTTP/2, HTTP/3 (QUIC) — Full header and body inspection for routing decisions.
• gRPC — Method-level routing with proto-aware metadata inspection.
• WebSocket — Session-aware routing with sticky affinity support.
• TCP/UDP passthrough — L4 routing based on source IP, port, and SNI.
• MQTT — IoT topic-based routing (Enterprise plan).

If your protocol is not listed, contact support@netfillx.com — custom adapters are available on Enterprise plans.

NetfillX uses usage-based pricing with three tiers:

• Free — 1M fill requests/month, all 6 services, community support. No credit card required.
• Pro — Starting at $49/month for 10M requests. Full analytics API, webhook delivery, email support.
• Enterprise — Custom volume pricing, dedicated edge capacity, 99.99% contractual SLA, named support engineer, and sub-1-hour incident response.

All paid plans bill monthly with no annual lock-in. Requests beyond your plan's included volume are billed at $0.40 per additional 1M. No egress fees, no per-endpoint charges, no hidden costs.

Getting your first fill request routed takes about 5 minutes:

1. Sign up and grab your API key from the dashboard.
2. Install the SDK: npm install @netfillx/sdk (Node), pip install netfillx (Python), or go get github.com/netfillx/go-sdk (Go).
3. Define your endpoints and routing rules via the API or dashboard UI.
4. Issue your first fill request — the SDK handles authentication, edge selection, and retries automatically.

Full quickstart guides are in the API documentation.

Rate limits apply per API key and are enforced at the edge:

• Free — 200 req/s burst, 50 req/s sustained.
• Pro — 2,000 req/s burst, 500 req/s sustained.
• Enterprise — Custom limits, no shared burst pool.

When a rate limit is exceeded the API returns HTTP 429 Too Many Requests with a Retry-After header. Our SDKs handle automatic retry with exponential backoff by default.

Yes — on-premise and private-cloud deployments are available on the Enterprise plan. We offer two models:

• NetfillX Private Edge — We deploy and manage edge nodes inside your VPC or data center. Full data residency and network isolation with the complete platform feature set.
• NetfillX Embedded Router — A container image of the core routing engine you run yourself. Receives rule updates from the cloud control plane but processes all fill requests locally. Suitable for air-gapped environments.

Both options are covered by the same contractual 99.99% SLA. Contact our enterprise team to discuss your architecture.

NetfillX measures and reports two distinct latency components:

• Routing decision latency — From when the edge node receives the fill request to when it issues a routing decision. Our global p50 is currently 0.78ms.
• End-to-end latency — Includes client-to-edge network transit plus routing decision time. Typically under 5ms for clients within 500km of a PoP.

Both metrics are available in real time in the Analytics Dashboard, broken down by region, endpoint, and percentile (p50, p75, p95, p99). Historical data is queryable via the Metrics API with millisecond granularity going back 90 days.

We deliberately use percentile-based measurements rather than mean latency, which masks tail behavior.

SLA commitments by plan:

• Free & Pro — 99.9% monthly uptime (best-effort, no financial credits).
• Enterprise — 99.99% monthly uptime with financial service credits. SLA breach credits: 10% of monthly bill for 99.9%–99.99% uptime, up to 30% for below 99.0%.

Uptime is measured as the percentage of minutes in a calendar month during which the routing API is available. Scheduled maintenance windows (announced 72 hours in advance) are excluded from uptime calculations.

Current uptime and incident history are always at status.netfillx.com.

All NetfillX API requests are authenticated using Bearer tokens in the Authorization header:

Your API key is provisioned at sign-up and available in Settings → API Keys. Key types:

• nfx_live_sk_ — Live key with full read/write access.
• nfx_test_sk_ — Sandbox key, no quota consumed.
• nfx_ro_sk_ — Read-only key for dashboards and monitoring.

Keys can be scoped to specific IP ranges, endpoint groups, or API namespaces. Short-lived tokens are available via the token exchange endpoint for client-side usage. All API traffic uses TLS 1.3; keys at rest are AES-256 encrypted. Enterprise plans additionally support mTLS and SAML-based SSO.