Compare two NTP servers — live measurement

Two servers, 3 SNTP requests each, in alternating order drawn at random and 500 ms apart. Each shot appears as soon as it comes back. We keep the sample with the lowest round trip, the one least affected by queueing delay. The second field offers ntp-8.rdem-systems.com, one of our own servers: it serves as a reference point, and it is measured exactly like the other.

What becomes of your query: if both servers answer and their names resolve to the addresses queried, the pair and its difference join the public table below. We record nothing about you — no address, no browser, no session — and private addresses are refused.

Latest measured comparisons

The last 20 pairs queried from this page, one row per pair. The difference compares the two servers with each other. We do not publish here each server's offset against our own clock, nor the round-trip times: those values describe our measuring infrastructure, not the servers queried. They do appear in the result of YOUR comparison, just above.

These rows are one-off measurements, taken from our networks at the time shown. They say nothing about a server's lasting quality: that needs a series, which is what the observatory does.

Further reading

• The measured French time servers — the inventory, server by server, with what was obtained and not what is claimed.
• The European NTP/NTS observatory — the full campaign, its sealed data and its method.
• The worldwide NTP and NTS map — by continent and by country, campaign after campaign.
• NTP strata explained — why a server's stratum does not tell you its quality.
• NTS, authenticated time — what a server must actually serve before we call it “NTS”.
• The state of our NTS pool — our own servers, measured continuously, with their offsets and certificates.
• Check your machine's clock — measurement the other way round: your clock, seen from here.
• The leap second — and how chrony handles it, from smear to step.