-DAVE MILLS STORIES
From Hans-Werner Braun, chief engineer of the NSFNET backbone
Vint,
I am really, really sorry to hear that. Dave had taught me how the Internet works. Eric Aupperle once, in perhaps 1984 or 1985 at the University of Michigan, gave me a couple of 8" floppy disks and asked whether I would be interested in checking them out. They were from Dave Mills with this RT-11-like Fuzzball code. I was familiar with DEC’s RT-11 from my prior work at the RRZK at the University of Cologne. I got hooked quickly on the Fuzzball software, this Internet concept was certainly much more enticing than working on code for Merit’s non-IP packet switches. I spent many long hours and evenings at the UofM Computing Center, working on the Fuzzball stuff and learning how the Internet works, via a 1200 bps modem. Dave Mills was living near the east coast then, though he may have initially been in San Diego before he went east, not sure. Nevertheless, Dave was usually just a phone call or an email away, and he seemed to be working day and night, usually responding very quickly. I have Dave to thank for mentoring me on Internet things, he was a good teacher, enabling various other things I did later. We also communicated a lot during the time of the 56 kbps NSFNET backbone. Prior to the NSFNET working, it suddenly became clear that we did not have an (assembly code) driver for the specific Ethernet cards that had been purchased for the NSFNET backbone. I then wrote the driver, using Dave’s existing Ethernet driver(s) as a template, and his time to help when I ran into issues. Dave invited me to the last GADS meeting, where during the meeting the GADS was split into the IETF and the IRTF (or ineng and inarc or whatever it was called initially). I continued to attend, and soon present, at subsequent IETF meetings.
I always had great respect for Dave Mills, and he deserved it. I have lots to thank him for, as his collaboration and mentorship greatly helped me to understand how the Internet was put together back then, and how it was working.
I wonder whether all around the world the Time Keepers dropped a second upon realizing that their Time Master had died?
Hans-Werner
===============
From Louis Mamakos, a key engineer of the pioneering UUNET
Dave Mills was one of the few people that directly changed the course
of my life, sending it well and truly off towards this Internet thing.
He taught a "Special Topics in Networking" class at the University of
Maryland during the time he was at COMSAT, in 1980 that I took from
him. The term project was bringing up a Fuzzball on a PDP-11/40 in
the student lab. And though various means, coming up with a 1200 baud
dial-up modem which we used to dial-up one of his Fuzzball and there
we were, on the early Internet in 1980.
The next semester was an independent study course with Dave, where
Mike Petry and I wrote a TCP/IP stack for the UNIVAC 1108 mainframe
computer we had. What great fun to get that on the end of our
1200 bps dial-up. Even taught it the HELLO protocol, too.
Later, Dave was the reason for my unhealthy fascination with timekeeping
and my early work with NTP, culminating with building the first UNIX
NTP implementation. It was a means of testing the completeness of
the NTP RFC draft to see if an implementation could be done from it.
(With out, of course, cheating and looking at the Fuzzball implementation.)
Thanks, Dave, for being so generous with your time and pointing me
in the right direction. You'll be missed by so many.
—----------------------------------------------------------------------------------------------
One other silly thing that reminds me of Dave's jolly nature and sense
of humor.. You might not know, but Dave was an amateur radio operator,
a hobby that I share as well. His amateur radio call was W3HCF -- I
couldn't decide if this was the luck of the draw, or he obtained a
"vanity" callsign from the FCC when he moved to Maryland.
"HCF" of course reminded me of the apocryphal "Halt and Catch Fire"
instruction for whatever mainframe computer we happened to be using
back in the day.
A few years after the course I took from him as a "visiting professor",
he was leaving COMSAT? Or it might have been MA/COM Linkabit.. I would
imagine the commute was pretty painful for his wife as he was legally
blind at the time and couldn't drive. He wanted to join the faculty
at U of MD as it was nearby and wouldn't have required a move. Mike
Petry and I brought this to Glenn Ricart, who was director of the
Computer Center at the time. He was all for bringing Dave on as
he came fully equipped with his own funding via DARPA and presumably
other sources. There was already a "Computer Vision Lab" associated
with the Computer Center that was largely self-supporting as well.
Of course, there was a pissing-match with the Computer Science Department
who insisted that research only happens over there. This turned
into the usual academic turf-war that drug on for too long. And then
UDel came along and I believe offered him a tenured faculty position
and off he went. What a huge loss that was for us at UMD, and gain
for UDel. What could have been, and all that..
He was quite the character.
I recall that first "Special Topics in Networking" class he taught in
the evenings, maybe two nights a week. First class we show up and
Dave explains how this is going to go. Mid-term exam, final exam
and semester-long project, each counting for a third of the grade. And
he expected something more than a book report for this final project.
It was clear at this point that we were in for an interesting ride,
and it wasn't the usual sort of class many had come to expect. Maybe
a quarter of the class dropped over the next couple of weeks when it
became clear some real effort was required.
Come time for the mid-term, Dave asks the class "So, who wants to
take the mid-term exam?" "No, we don't need no stinking mid-term!"
was the prevailing opinion. "OK, so no mid-term!" I'm sure this
was a real time-saver! And he reminded us that now our grades
relied on the final exam and project, for which he had high expectations.
I think this all played out before the final drop-date for the class,
and some number of people decided the odds didn't look great.
Meanwhile, there was no text book for this class. There's IEN documents
and some early RFCs and some papers.. this is really contemporary
material. And, of course, Dave's name appears in more than a few of
these papers that we're assigned to read. Hmmm...
Mike Petry and I were in this class, and our project was bringing
up the Fuzzball software on this PDP-11/40 in the undergrad student
lab. We'd each sign-up for an hour, and we'd managed to get our
own RK05 disk pack with the RT-11 system that hosted the Fuzzball
code that we were building and running. It was a whole new,
wonderful world once it got working and our 1200 baud dial-up
connection to one of his Fuzzballs was up! PING and FTP and TELNET!
Wow!
Come time for the final, Dave once again asks who want's to take a
final. This time, it was optional. Many people decided to hedge
their bets, take the final and not rely solely on the Project they
should have been working on all semester.
There seemed to be either disappointed people or really happy people
afterwards. Sure, we got a good grade (probably an "A"), but what a
unique, memorable and literally life-changing experience that was for
me. This was the REAL DEAL and you could tell this was something
very special we'd learned about, delivered by someone that was
clearly very passionate about what he was sharing with us.
It was a wonderful experience, and I'm so fortunate to have the
luck to be at the right place, at the right time to have Dave Mills
as a teacher and mentor.
louie
==============================================
From Karl Auerbach, early Internet pioneer
Oh my, We have lost Father Time.
We have all - everyone has - been the beneficiary of his ideas and work.
===============================================
From David Täht, “Mr. Bufferbloat”
oh, darn! I was just thinking about him the other day. I thought that
our industry should give him a giant gold pocketwatch for his
retirement, synced to an atomic clock....
Never met him. Would have loved to.
RIP
====================================================
From Leigh (nee Mills) Schnitzler:
Here’s a couple of my memories of Dad. I’m not positive of the exact dates, but I think they were all somewhere around 1975-1977…
Keith and I were curious as to what he did with all the time spent in his den. At the time he was working with ISI (?) out in California and they had a server we called ISI-E. He taught us how to pick up the phone, call that server, then put the receiver into a coupler which was sitting next to the phone. This was connected to a line printer at first, then later to a green-screen terminal. Once we had a command prompt, we would start up Zork. If it was after 5pm west coast time, we would get “You are standing in an open field west of a white house, with a boarded front door. There is a small mailbox here.” If it was before 5pm, a knight in black (?) armor would come out of the woods and chop your head off with a big sword.
In regard to his being an Ass-Prof (as he called it) at UofM - One of Dad’s students had atrocious skills in the English language. As you probably know, Dad was a stickler for the proper use of the language (even if it was in Mills-speak!!) He handed me a term paper, along with a red pen, and asked me to mark it up for grammar and punctuation. Wow, did I ever! There/their/they’re, to/too/two, your/you’re….. Missing commas and periods…. It was full of red marks! He then took it to I-don’t-know-who (the Dean, maybe?) and said that his 9-year-old had graded it for grammar and punctuation. He didn’t think this student should be permitted to graduate unless and until he had a better understanding of the written form of the English language. The reason was that it would reflect poorly on the UofM as a teaching institution. The response he got back was along the lines of “you are not in the English department, so you do not have a say in this”. The student was permitted to graduate.
Last but not least: He also taught me a little bit of Basic code. I was a brat in 7th grade, and was cursing at somebody in music class, so the teacher told me to write “I will not say Jesus Christ in class” 100 times and turn it in the next day (you can see where this is going!) To the computer I went!
10 For A=1 to 100
20 Print “I will not say Jesus Christ in class.”
30 Next A
40 End
I turned in a multi-page printout the next day, and the look on the face of the teacher was priceless! His response was “I’ll let you get away with it this time as you had to learn the computer code to do this. Next time though, you need to write it out.”
Leigh is David Mills daughter
=====================================================
UDel’s Mills Computer Cluster
https://www.hpc.udel.edu/systems/mills
=====================================================
A lecture by David Mills:
It’s titled “A Maze of Twisty, Turney Passages - Routing in the Internet Swamp” and can be found here: https://www.youtube.com/watch?v=08jBmCvxkv4
From Leigh Schnitzler
=======================================================
From Barbara Denny:
I believe I first met Dave Mills when he was at Linkabit in the early 80s. Even though I also think of him as Father Time, his Fuzzballs were a frequent topic at meetings I attended. In checking my Linkabit memory, I found this oral history. I always enjoyed meetings more when he was present.
https://conservancy.umn.edu/handle/11299/113899
========================================================
From Ole Jacobsen (author/editor of Connexions and Internet Protocol Journal)
Sorry to hear that. Our community has indeed lost a very special person.
My memories of Dave go back to my days when I was working at the Norwegian Telecommunications Administration Research Establishment (NTARE) with Paal Spilling. Dave had a Fuzzball router at NTARE and would occasionally send us instructions and pointers to patches he had made while were all asleep. His use of language always amused me: "I left on your Fuzzball a flock of files..."
He will be missed!
Ole
===================================================================
From Jack Haverty, an early Internet implementer and pioneer:
In the infancy of The Internet during the early 1980s, Dave was my
friend and colleague. He was always curious, competent, innovative,
informative, and playful - for example, I learned from Dave that the
popular PING tool was actually an acronym: Packet InterNet Groper.
Internet meetings and mailing list debates were fun when Dave was involved.
Dave was also my nemesis. As the explorer and tinkerer scientist, Dave
was always trying out new ideas for his potent army of Fuzzies - the
small computers he was seeding around the Internet. At the time, my
task from Arpa was to "make the core Internet run as a reliable
service". Tinkering and Reliability do not mix well.
Dave was the Scientist of the Internet. At Internet meetings, Dave
would report on his latest Fuzzy Adventures. I can paraphrase his
reports as "I tickled the Internet with a Fuzzy stick and it turned
pink!" My reaction was always "Don't do that!!"
Few people probably realize one of Dave's crucial contributions to
Internet technology - the concept of "Autonomous Systems", which
survives today, 40 years later. Dave didn't invent it, but he caused
it. The antics of his Fuzzies forced Eric Rosen and I to create
Autonomous Systems and EGP as a "firewall" mechanism to keep the
Internet "core" safe from the marauding Fuzzies and their occasional
collateral damage.
Scientists yearn to measure things. Engineers just want to make it
work. In the 1980s, the scientist part of Dave needed to measure time
on the Internet and there was no way to do that. So he created one,
quite elegant and unbelievably accurate. Then his Engineering persona
made it work.
The Internet will never forget Dave. As the Internet evolves into the
Internet of Things, and billions of devices continue to appear, they
will all know what time it is, thanks to Dave.
Dave is still here, all over The Internet.
==========================================
From Paul Vixie, another early Internaut and founder of the Internet Systems Consortium
My first assigned task at DECWRL when I arrived in 1988 was to "upgrade the fuzzball". Learning what a fuzzball was took me to Dave Mills who helped me figure out out and eventually telnetted into the new box to rebuild the kinemetrics truetime driver on the later RT11 kernel. Later he explained to me all about the Hello protocol. Still later he let me rewrite the Unix driver for the kinemetrics clock. I still have the clock and the PDP11 fuzzball once named clepsydra.dec.com in my basement. I never met anybody else quite like Dave and I don't expect I ever shall.
p vixie
==========================================
From Peter Allen (former student):
My meager contribution to the collection of memories that you are assembling in memory of Dave Mills …
I first met Dave as a senior Computer Science student at University of Maryland. Having taken several introductory courses in operations research, graph theory, and basics of computer communications, my advisor suggested that I take Dave’s evening course on advanced networking concepts.
Dave was unlike any professor I had ever encountered. He traveled from Delaware for these courses (not an easy transit given his vision challenges) and came prepared with enthusiasm and energy that epitomized genuine passion for the potential of autonomous operation of computing units. His use of colloquialisms and anecdotes conveyed the immense potential that he envisioned. He challenged his students to think differently and to imagine a world order that broke virtually every commonly accepted constraint.
I had worked at Mitre as a co-op student and had concluded that coding was not my forte. Dave helped me to see a different career trajectory with my Computer Science degree and inquisitive mind. He cared. He broke barriers to thinking and shared a sense of immense potential from interoperable networking connections.
While Dave was a part-time instructor, my tenured professors held Dave in high regard and reinforced his coaching. On graduation I sought a position with BBN to be involved with ARPANET and emerging NCP-to-TCP transition that was core to internetworking at scale.
I went on to join the team at BBN, supporting DCA for DDN, the intelligence community, and even MCI (MCI-MAIL)! None of this would have happened for me without Dave’s coaching and energetic enthusiasm. To be clear, I was not superlative. I was ordinary. Dave lit my fuse. Rather than slog away at COBOL programming (for which I would have been horrible), I followed a passion encouraged by a man who cared about people as much as he did about fuzzball innovations.
Rest, Dave. You did well.
================================================
From Ajit Thyagarajan, PhD student of Dave Mills
Dave played an outsized role in shaping my career by not only introducing me to the incredible world of the Internet, but also in helping make all the connections needed to be successful in the field.
I remember when I first joined UD and started my work with him, his lab had all these funny named machines. He then went on to give me a tour of the Okeefenokee swamp and told me all about Pogo, Malarkey, Beauregard and others in great detail. They were all characters from Walt Kelly's comic strips back in the 1930s and what a trip it was. And since my name started with A, I managed to get Albert as my machine.
Dave was a "geeky collector". Louis Mamakos mentioned that he got W3HCF as his ham radio call sign which was cool. He got port 123 for NTP - super easy to remember. And he had an entire Class B network for the lab and it was 128.4.0.0/16 (did you notice the .4?)
I remember in my second week of being at UD, Dave sent me an abstract for a seminar series I was doing in the Department. I looked at the abstract and found it to be riddled with grammatical errors and words that were not English. Thinking that he might have hastily typed it up, I proceeded to correct the abstract and send it out to the Department. The next day, Dave summoned me to his office and reprimanded me for correcting his words. That’s when I got the lesson on Millspeak!
I was very fortunate to get to know him and Barbara personally over the years. He will always be in my memories. RIP!
================================================
The following are the words of Poul-Henning Kamp (https://phk.freebsd.dk/). Originally published at the blog:
https://www.version2.dk/holdning/rip-dave-mills
Current form is a translation from danish to english by Google and Claus Andersen. I have notified PHK that this exists - he may follow up himself.
--------------------------------------------------
RIP: Dave Mills
Time and radio time signals have always fascinated me.
As approx. 7-8 years old inherited an old "steam radio" that could receive the VLF band and there I could hear both "Rugby" on 60kHz and "DCF77" on 77.5 kHz: beep-beep-beep-beeeeeep-beep-beeeeep...
I was a network engineer at DKUUG for a short period when we switched from UUCP to TCP/IP and of course I experimented with NTP, but our Sun 3/50 machine had a kernel error which made it not run very well.
While I was in Luxembourg I found diagrams and a description of an ISA card with a Loran-C receiver[1] on the University of Delaware's FTP server and I thought about it quite a lot, but my soldering iron was at home in Denmark, so it only became a dream.
While at FLS data I had a "skunkworks" project to get our computers to just roughly agree on what time it was and found that if you configure a serial port to 25 bps it can receive the DCF77 signal directly. I posted to the UseNet group comp.protocol.ntpabout the trick. (Many years later, a shady German tried to save a patent application by claiming that he had, out of the blue, sent an email to my boss at the time (Hi Jette!) offering just that solution.)
One day I spotted in one of these small 5x5cm product announcements in the back of Ingeniøren that Motorola had made a GPS receiver for time purposes, "Oncore UT". I contacted the Danish importer and they procured an "eval kit" for a small fortune.
That was the start of my friendship with Dave Mills and together we led NTP down into the dark world of nanoseconds and of all my publications I will always be most proud of the small article that Dave and I published about it[2] . My Erdös number is four, but my Mills number is one!
One of the funny things that came out of it was a skeptical phone call from a lady from the Air Traffic Control Service: She had contacted Dave to find out who could help them with an SNMP managed NTP server when they had to digitize in a few years the entire aviation security in Denmark. Dave had laughingly replied "I know just the guy: call 53 53 12 24"
At one point, Dave tried to make me his successor on the NTP project and he asked his friends in DoD and DARPA to find money so that I wouldn't have to do it as a hobby. To his, but not my, astonishment, they declined. Personally, I don't think they were half as interested in NTP as they were in Dave.
At one point my world exploded due to my ex-wife's serious mental illness: From Wednesday night to Thursday morning I became self-employed to take care of our two small children and I swam against the current as best I could.
After a long period of radio silence, I sent an apology to Dave, explaining what was going on and that I had needed to do something "100% useless". His response was a cryptic "Send me your address, I have just the thing."
A little over a week later, a package arrived from the University of Delaware, containing some opaquely serious paperwork about lending DoD sponsored hardware to a NATO country and a copy of the Loran-C receiver I had dreamed of ten years earlier.
It turned out to be the "one and only" serial number because Dave and his students had never really gotten the software to work…
Challenge accepted!
I've never been very reverent about my own abilities when it comes to making computers do weird things, but this was a project where failure is not an option.
During the day, I juggled the children to and from kindergarten, worked for clients, worked on FreeBSD, fought with the authorities and ex-wife over visitation.
But after the bedtime story was read I made a big cup of tea, took a deep breath, and went out into my tiny lab/engine room in the garage and wrestled with Dave's Loran-C receiver.
(Sorry for the blurry photo, it was taken with the heroically lousy Leica DigiLux[3] camera)
I have no idea how many thousands of hours I spent on that receiver before I could send Dave this blurry photo:
I'll spare you the explanation of exactly what those dots mean, it hardly ever interested anyone but Dave and me, but I was proud as a pope and Dave was proud as a new father.
Later I built two generations[4] of SDR Loran-C receivers[5] and Dave was over the moon when I sent him this little dancing pulse:
Animated version found here: https://phk.freebsd.dk/AducLoran/animation2.gif
Dave accomplished many things in life, he started his career with the "Bandits of the Beltway" as they were called in "MilesSpeak" and there were things he was still not allowed to talk about, for example the guidance computer of the Minuteman-1 missiles, things no one wanted to hear on, e.g. 16-pulse Tactical Loran-C and things that are so big that they are hard to grasp: If Vint Cerf was the "father of the internet" Dave Mills was its grandfather.
Somewhere in the middle of that long list should also be said that he kept a single father and two young children sane and on track, on the opposite side of the globe.
Glory be to his memory!
Links:
[1] https://www.eecis.udel.edu/~mills/database/reports/loran/
[2] https://www.eecis.udel.edu/~mills/database/papers/nano/nano.pdf
[3] https://kylecassidy.com/projects/dakota/about.html
[4] https://phk.freebsd.dk/loran-c/
[5] https://phk.freebsd.dk/AducLoran/
==================================================
From Ars Technica:
==================================================
From Daniel Pezely (a former student):
Mills’ Principles:
Guidelines used while developing the Internet Protocol stack, as conveyed by Dave Mills (a.k.a., Mr. NTP himself, creator of fuzzballs, original IP routers)
- You cannot anticipate all the faults.
- All fault scenarios will happen at least once.
- No one single strategy will work.
- The system must be self-correcting.
- But each correction must not increase vulnerability.
- No system will always obey these rules.
Amendment:
- No system works correctly.
Acid test:
Would you trust your paycheck to the system?
Implementation Hint:
Use unique timeouts such that even in combination they become signatures of where problems may exist. Use a different prime number for each to aid in discovering the combination of compounded timeouts.
===============================================
From Cathy Aronson:
“ I just found out that Dave Mills died in January. Another loss for our community.
I have a funny story about Dave Mills. In 1988 when I worked on the NSFNet someone shot (with a gun) at the fiber optic cable to Sesquinet and took it offline. There was no redundancy at that time. I had this vision of good old boys in the bayou in a row boat shooting at the only thing that wasn’t moving, the cable. I wrote an email about the good old boys in the bayou shooting at the fiber optic cable. My boss at the time, Hans-Werner Braun, sent out my email in a regional techs update email. I was a little embarrassed but it is what it is. Anyway I went to my first Interop in 1989 and met some guys from the Canadian version of the NSFNet. One of my co-workers had heard from them about Dave Mill’s talk and she told them to tell me about it.
The Canadians found me and told me this story.. Dave Mills had a section in his talk about hazards in networking. He called it the GOB factor… How many Good Old Boys it takes to partition your network. I met him later during the meeting and we both laughed about the GOB factor. I got to explain to the Canadians about my email and Sesquinet and how the GOB factor came to be.
Rest In Peace Dave”
========================