protocollen · 9 min read

How to choose an NTP or PTP time server: a decision guide

Which time server fits your application? This decision guide walks you through protocol, time source, holdover, form factor and sector in five steps.

Choose your reading level

Choosing a time server comes down to five questions. How accurate must your time be, where does the time come from, what happens if that source fails, which form factor fits your installation, and which sector do you work in?

Accuracy: milliseconds are usually enough (NTP). If you need microseconds, as in financial trading or a high-voltage substation, PTP comes into play.
Time source: almost every server derives its time from GPS/GNSS via a rooftop antenna.
Fallback (holdover): if GPS drops out, the server keeps running on its own oscillator. The longer you want to bridge that flawlessly, the better (and more expensive) that oscillator.
Form factor: a compact desktop, a 19-inch rack module, or a grandmaster with redundant power.
Sector: data centre, defence, broadcast and industry each have their own requirements.

Not sure? That is normal. Tell us your application and we will point you to the right one. Contact us.

Choosing the right time server is a matter of five decisions: protocol, time source, holdover, form factor and sector. This guide walks you through them step by step and links to the in-depth articles.

Step 1 - Protocol: NTP or PTP?

This is the first and biggest choice. NTP achieves millisecond accuracy on an ordinary network; PTP (IEEE 1588) achieves sub-microsecond, but requires PTP-aware switches.

Needed for	Choose
Servers, logging, clocks, IT	NTP
Financial trading (MiFID II), digital substation (IEC 61850), 5G, frame-accurate broadcast	PTP

In depth: NTP vs PTP: when to choose which protocol.

Step 2 - Time source: where does the time come from?

Most servers discipline to GPS/GNSS via a rooftop antenna. In an environment without an antenna, the time can also come from an existing IRIG or PTP source.

In depth: GNSS disciplining and stratum levels.

Step 3 - Holdover: how long without GPS?

If the GNSS feed drops out, the server keeps running on its internal oscillator. The type determines the autonomy: TCXO (minutes to hours), OCXO (days), Rubidium (months). Choose based on how bad drift is and how quickly you can restore GPS.

In depth: Holdover: TCXO, OCXO and Rubidium.

Step 4 - Form factor and redundancy

Compact / desktop for a single location or test bench.
19-inch rack module for a data centre or server room.
Grandmaster with redundant power and dual GNSS for critical, traceable time.

In depth: Reliable time: redundancy and traceability.

Step 5 - Sector

Every sector has its own requirements. Look at the application closest to you:

Which Masterclock model fits?

Compare the families directly:

NTP100-GNSS vs -OSC vs -TC - compact NTP servers.
GMR1000 vs GMR5000 vs GMR6000 - grandmasters.
NTP time server vs GMR grandmaster - the core choice.

Still unsure? Tell us your application and we will recommend the right model.

Choose a time server from the error budget, not from the product brochure. First determine the required accuracy relative to UTC, the traceability, the holdover requirement and the redundancy; those four determine protocol, oscillator and form factor.

1. Error budget (accuracy)

Set the maximum permissible deviation from UTC at the point that matters (the application, not the server output). NTP typically delivers 1-10 ms over a LAN; PTP delivers 100 ns - 1 us, provided PTP-aware switches (boundary/transparent clocks) manage path asymmetry. Account for the full path, not just the source.

2. Traceability

For auditable time (MiFID II RTS 25, IEC 61850), the chain must be traceable to UTC through a known stratum structure. Document the source (GNSS, disciplined oscillator), the stratum and the measurement uncertainty.

3. Holdover requirement

Holdover is the integral of the frequency error: over a long outage, drift dominates, not the initial offset. Determine the requirement as (permissible deviation) divided by (source recovery time). An OCXO that was well disciplined just before the outage performs above its nominal specification. See holdover sources.

4. Redundancy

Critical installations duplicate the time source (two independent GNSS antennas, or GNSS + IRIG/PTP), the power supply and the network path, with automatic failover. See reliable time.

From requirement to model

Profile	Direction
ms accuracy, IT/data centre	NTP100-GNSS
ms with long holdover	NTP100-OSC (OCXO)
sub-us, PTP grandmaster	GMR5000 / GMR6000
Power Profile (IEC 61850-9-3, C37.238)	GMR6000

Sources: IETF RFC 5905 (NTP); IEEE 1588-2019 (PTP); NIST Time & Frequency Division; ESMA RTS 25 (MiFID II).