Time Server Buyer's Guide: 10 Must-Ask Questions
Essential questions for selecting precision timing systems
Time Server Buyer's Guide
10 Must-Ask Questions Before You Invest in Precision Timing Infrastructure
2025 Edition · Actionable Checklist InsideIntroduction — Why Timing Matters
In an era of 5G fronthaul networks, distributed financial trading platforms, synchronized smart grids, and defense-grade sensor fusion, time is not just a metric — it is the invisible backbone of every mission-critical system. A deviation of even a few microseconds can translate into dropped calls, regulatory fines, misaligned radar returns, or a complete loss of situational awareness.
Modern time servers — also called Network Time Servers, PTP Grandmasters, or GNSS-disciplined clocks — sit at the heart of this infrastructure. They ingest satellite signals (GPS, Galileo, GLONASS, BeiDou) or atomic references and distribute traceable time across enterprise and carrier networks using protocols like NTP and PTP (IEEE 1588).
Choosing the right time server is a multi-dimensional decision that spans accuracy, protocol support, resilience, environmental tolerance, integration complexity, and total cost of ownership. This guide distills the process into 10 essential questions you must answer before issuing an RFP or placing a purchase order. Each question includes a practical checklist you can bring into your next vendor meeting.
1 What Accuracy Do You Need?
Accuracy requirements vary by orders of magnitude depending on the application. A campus NTP deployment may be perfectly served with ±1 ms, while a 5G TDD network demands sub-microsecond alignment, and scientific or defense applications may require nanosecond-level precision traceable to UTC.
| Application | Typical Accuracy | Resolution |
|---|---|---|
| Enterprise IT / NTP | ±1 – 10 ms | Milliseconds |
| Financial Timestamping (MiFID II / Reg NMS) | ±1 µs | Microseconds |
| 5G Fronthaul / TDD Sync | ±1.5 µs | Microseconds |
| Power Grid (IEEE C37.238) | ±1 µs | Microseconds |
| Defense / SIGINT | ±10 – 100 ns | Nanoseconds |
| Scientific / Radio Astronomy | ±1 – 10 ns | Nanoseconds |
✅ Accuracy Checklist
- Define the accuracy your primary application demands (µs vs. ns)
- Account for end-to-end error budget: time server output + network asymmetry + client oscillator
- Determine whether you need traceability to UTC (NIST, NPL, PTB) with documented uncertainty
- Verify the vendor's stated accuracy is measured at the output, not just internally
- Ask for third-party test reports (e.g., NIST-traceable calibration certificates)
- Consider future-proofing — will your accuracy needs tighten within 3–5 years?
2 What Synchronization Protocol?
Network Time Protocol (NTP, RFC 5905) and Precision Time Protocol (PTP, IEEE 1588-2019) serve fundamentally different precision tiers. Many modern deployments require both: NTP for legacy IT assets and PTP for latency-sensitive applications.
| Aspect | NTP | PTP (IEEE 1588) |
|---|---|---|
| Achievable Accuracy | 1 – 10 ms over WAN | < 1 µs (hardware timestamping) |
| Transport | UDP port 123 | Layer 2 (EthereType 0x88F7) or Layer 3 UDP |
| Profiles | N/A | Default, Telecom (G.8275.1/.2), Power (C37.238), Media (SMPTE ST 2059) |
| Hardware Timestamping | Optional (usually OS-based) | Required for sub-µs performance |
| Scalability | Very high (stateless UDP) | Moderate (boundary clocks recommended) |
✅ Protocol Checklist
- Confirm NTP version support (NTPv4 recommended; SNTPv4 acceptable for basic use)
- Confirm PTP profile support — Default, IEEE C37.238 (power), G.8275.1 / G.8275.2 (telecom), SMPTE ST 2059 (broadcast)
- Ask about multi-profile simultaneous operation
- Verify support for both Layer 2 and Layer 3 PTP transport
- Confirm hardware timestamping on all PTP-enabled Ethernet ports
- Determine maximum PTP client scale per port and per chassis
- Confirm multicast and unicast PTP message modes
3 How Much Holdover Do You Need?
"Holdover" describes the ability of a time server to maintain accurate time when its primary reference (GNSS signal or external atomic clock) is lost. During a GNSS outage — caused by antenna failure, jamming, spoofing, solar interference, or construction — the internal oscillator free-runs, and accuracy degrades over time at a rate determined by the oscillator quality.
| Oscillator Type | Typical Drift | Holdover Performance |
|---|---|---|
| TCXO (Temperature-Compensated) | ±0.5 – 5 ppm | Seconds of drift within hours |
| OCXO (Oven-Controlled) | ±0.01 – 0.1 ppm | ±1 µs over 24–72 hours |
| Rubidium Atomic | ±0.001 ppm (1 × 10⁻¹²) |