David Mills, an Internet Pioneer, Has Died 19
David Mills, the man who invented NTP and wrote the implementation, has passed away. He also created the Fuzzballs and EGP, and helped make global-scale internetworking possible. Vint Cerf, sharing the news on the Internet Society mail group: His daughter, Leigh, just sent me the news that Dave passed away peacefully on January 17, 2024. He was such an iconic element of the early Internet.
Network Time Protocol, the Fuzzball routers of the early NSFNET, INARG taskforce lead, COMSAT Labs and University of Delaware and so much more.
R.I.P.
Network Time Protocol, the Fuzzball routers of the early NSFNET, INARG taskforce lead, COMSAT Labs and University of Delaware and so much more.
R.I.P.
Well, it was his time. (Score:5, Funny)
Re: (Score:1)
Too soon? :-)
Nonsense. Quite timely.
Re: (Score:3)
Re: Well, it was his time. (Score:1)
Re: (Score:3)
But seriously, with Father Time dead who's going to keep NTP development going?
This is actually a very serious problem. NTP is both mature and stable, so it will be fine for decades, but not indefinitely. Once our infrastructure sufficiently changes that some of the underlying assumptions in NTP are no longer true, someone will have to do digital archeology to try to address these. That is, if everything goes well. There is a still a small possibility of a critical CVE coming out. Then what? We stop synchronizing servers?
Re: (Score:3)
My condolences to his family.
OpenBSD forked ntpd and created OpenNTPD. So people can always move to OpenNTPD if they are concerned. But I thought a foundation or group of people took it over a couple of years ago.
https://www.openbsd.org/openntpd/features.html
Re: (Score:2)
I've not used it but there is also NTPSec: https://ntpsec.org/ [ntpsec.org]
Re: (Score:1)
OpenNTPD wasn't a fork of ntpd, it was written from scratch. It is also not a direct replacement for ntpd, as it only serves up "good enough" time. The design decisions, security over time accuracy, are well documented.
Re:Well, it was his time. (Score:4, Informative)
Mills hasn't been in charge of the spec or the reference client since 2010. Both are maintained by others now, with the protocol being an official one handled by the IETF.
Re: (Score:3)
The NTF maintains the reference implementation of NTP. As I understand it, they're not as well funded as some of the other Open Source projects. This is really one of those cases where the XKCD cartoon really gets it right. https://xkcd.com/2347/ [xkcd.com]
Last I had heard, Harlan Stenn, one of Dr. Mills assistants was still involved. The foundation is currently fundraising to acquire some SSD's. You can donate here: https://www.nwtime.org/ssd-app... [nwtime.org]
Harlan is indeed in charge (Score:5, Interesting)
I've been a member of the NTP Hackers group of maintainers since the 1990'ies, over most of that time Harlan has been in charge, with close to zero resources. The only thing xkcd got wrong is that Harlan moves up and won along the US west coast, instead of Nebraska.
When Covid happened my then job went away and until I found a new one (at Cognite, a Norway-based startup), Harlan tried to come up with funding for me to move into fulltime NTP development.
I had previously written a high-perf multi-threaded version of the stock ntpd deamon, my branch version was capable of handling maybe 100 M users all polling at the minimum 64 s poll interval.
Terje
Re: (Score:2)
David Mills, an Internet Pioneer, Has Died
Netcraft confirms it?
Re: (Score:2)
He just went to Stratum 15, reachability 000...
Re: (Score:2)
Too soon? :-)
His time ran out.
Stop all the clocks (Score:2)
Obligatory XKCD (Score:4, Funny)
He, better than perhaps any human that ever existed, would appreciate the joke.
Re: (Score:2)
Too soon.
Re: (Score:3)
Prof Mills have done a lot of work on this particular problem, i.e. Timekeeping in the Interplanetary Internet:
https://www.eecis.udel.edu/~mi... [udel.edu]
Terje