Clock Drift Explained: Why Network Devices Fall Out of Sync

Could the network fault you are chasing actually be a clock problem?

With a constant frequency error of just 12 parts per million, an uncorrected clock can gain or lose roughly one second a day, as the NTP Project explains.

Across a network, inconsistent time can obscure incident timelines and disrupt applications.

What is clock drift?

Clock drift is a clock gradually gaining or losing time relative to a reference. Small differences in the rate of local clock hardware accumulate unless corrected.

Clock offset is the difference at a particular moment. Monitoring offset over time helps distinguish gradual drift from sudden corrections or source changes.

Why do network devices fall out of sync?

Natural drift is one cause of incorrect time. Source, configuration and network problems can produce similar symptoms:

• Oscillator behaviour: temperature and ageing can alter clock rate.
• Lost synchronisation: an unavailable time source can leave a device relying on its local clock.
• Configuration: policies, blocked NTP traffic and fallback sources can affect synchronisation, as IBM explains.
• Virtualisation: host settings and pause or resume events can affect guest clocks. See Microsoft’s VM timing guidance.
• Network delay: unequal outward and return delays can bias NTP measurements, as NIST research demonstrates.

When does clock drift become a problem?

Clock differences can cause authentication failures and difficulty correlating logs. The acceptable error depends on the application: precise event ordering may require much tighter synchronisation than routine IT operations.

Does NTP fix clock drift?

NTP’s clock discipline algorithm reduces time and frequency errors. It still needs an appropriate reference, sound configuration and suitable network conditions.

For tighter requirements, our PTP vs NTP guide explains when PTP may be appropriate.

How to investigate clock drift

Check	What to establish
Active time source	Confirm which source each endpoint actually uses.
Offset over time	Check whether error grows steadily, jumps suddenly or follows a source change.
Timing path	Investigate packet loss, variable delay and unequal forward and return delays.
Host and guest behaviour	Check whether time changes coincide with VM pauses, resumes or host source changes.
Reference failure	Test how long endpoints stay within tolerance during controlled reference loss.

How edgeTime helps

edgeTime helps assess performance, review architecture and plan resilience through timing consultancy and timing and synchronisation solutions.

Frequently asked questions

Is clock drift the same as clock offset?

Offset is the time difference now. Drift describes a clock gaining or losing time relative to its reference.

Can servers drift even when NTP is enabled?

Yes. The service may be failing, the wrong source may be selected or accuracy may fall outside the application’s requirements.

Does manually setting the clock solve drift?

Only temporarily. Setting the time does not remove the underlying rate error; continued synchronisation is needed to control it.