On Thu, Feb 14, 2013 at 3:56 PM, Ask Bjørn Hansen <[email protected]> wrote: > > On Friday, February 8, 2013 at 7:50, AlbyVA wrote: > > Does anybody have any metrics on the market share of NTP traffic that > > goes to servers under ntp.org (http://ntp.org)? > > It really depends on what you mean by "global". > > time.apple.com and time.windows.com probably have the most share (by far?). > > Do you include private/internal servers? If so then > Google/Facebook/Rackspace/...'s share might be significant. > > Some years ago my estimates for our queries per second where comparable to > the queries per second that the NIST servers got at the time; so if you just > compare with other "generally public" services, I think a reasonable > guestimate would be 40-70%. >
Even if you could reliably sample all Internet NTP traffic, it still wouldn't tell you what's going on internally in organization networks. For example, the stratum-2s at all of my org's locations use *.pool.ntp.org as non-preferred "sanity check" servers, and then serve their time to hundreds of other internal devices if our stratum-1 sources are unavailable. Do all those hundreds of internal devices "count" as being served by the NTP pool project, even though there are only 8 of my servers actually querying the pool.ntp.org over the Internet? Similarly, I have seen what I know are pool servers in the RefID field at some cloud services on some ISPs; potentially millions of devices are getting time from pool.ntp.org indirectly. -- RPM _______________________________________________ pool mailing list [email protected] http://lists.ntp.org/listinfo/pool
