Service Level Agreement

Commitments at a glance

Uptime definition

"Uptime" means the percentage of minutes in a calendar month during which a Service is reachable and responsive from our edge. Specifically:

• Network uptime — ICMP + TCP:443 reachability from at least two external monitoring probes outside our ASN.
• Power uptime — power delivery to the server's PDU, measured by the facility's monitoring.

Downtime is measured in whole minutes, rounded up, starting at the first failed probe and ending at the first successful probe after restoration.

What does NOT count as downtime

• Scheduled maintenance announced at least 14 days in advance, published on /status, and capped at 2 hours per calendar month.
• Customer-caused outages — kernel panic from a customer-built kernel, OS misconfiguration, firewall blocking our probes, deliberate shutdown from the customer panel.
• Force majeure — earthquake, war, strike, government-ordered blackout, BGP hijack not originating from our ASN, submarine cable cut affecting all diverse carriers simultaneously.
• DDoS attacks exceeding 1 Tbps that saturate our scrubbing capacity — we commit to 1 Tbps of absorption; larger events are force-majeure.

All other downtime — including upstream carrier failures we could have routed around, DDoS within our committed capacity, power faults at the datacenter, and internal software bugs — does count.

Service credits

Missed-SLA credit is issued automatically on the following invoice as a line-item reduction of your monthly base fee (not including add-ons). You do not need to file a request.

Hardware-swap credit

If we miss the 4-hour hardware-swap SLA on a dedicated server (measured from confirmed diagnosis to online replacement), we credit one (1) additional day of base fee for each hour past the deadline, up to 100 %.

How we measure

We publish the raw monthly uptime per PoP at /status. External probes are operated by third parties not affiliated with us (StatusCake and UptimeRobot as of April 2026). All historical data is retained publicly for 5 years.

If your own monitoring disagrees with ours, send the data to [email protected]. We review within 5 business days and issue additional credit if your data is more accurate.

Support response targets

"First response" means a human engineer reading your ticket and replying with either a diagnosis or a request for clarification — not an auto-acknowledgement.

• Tier 1 (outage) — anything affecting reachability, availability or data integrity. 15-minute first response, 24/7.
• Tier 2 (degraded service) — performance issue, intermittent error, single-node problem. 4 business hours.
• Tier 3 (questions, how-to) — configuration advice, feature questions. 1 business day.

Claims procedure

We issue credits automatically for the metrics we measure. If you believe you are owed additional credit (private monitoring, hardware-swap time, etc.):

1. Email [email protected] within 30 days of the end of the affected billing period.
2. Include the server ID, the affected time window, and your measurement data.
3. We respond with a ruling within 10 business days. Credit, if owed, appears on the next invoice.

Service credits are the sole financial remedy for SLA breach — they do not limit our liability under the Terms of Service, but they are the exclusive mechanism for SLA-related compensation.