How we got the modern Internet
ฝัง
- เผยแพร่เมื่อ 29 พ.ค. 2024
- Ever wonder how we ended up with the Internet we have now ? What string of events got us to where we are now. I'm guessing there is a version of the history of the Internet you've been told, but was it true, or just a small part of the what happened.
This video is sponsored by PCBWay (www.pcbway.com)
My thanks also got to :-
Nostalgia Nerd
Stu
WhatHoSnorkers James O'Grady
Ed, Mike & Terry
0:00 - Introduction
0:33 - Time to bring in the big guns
1:03 - Voxpops (we are like a proper news show now)
1:58 - It all starts with a memo
2:35 - A word from our sponsor
2:53 - Arpanet
4:30 - Donald Davies, & packet switching (the Internet is really Welsh)
5:22 - BBN and the Imps
9:09 - Arpanet, and the NPL network join to form the Internet (yes the creation of the internet is a join UK US thing)
12:20 - TCP/IP
15:17 - Class-full routing (we still don't have the modern internet)
17:15 - CIDR & BGP (the modern internet is upon us)
18:43 - Arpanet shutdown, and the rise of new networks
20:03 - The administration structure of the internet changes
20:41 - The road to DNS
24:43 - The changing culture of the internet
27:30 - IPv6 and the future of the internet
32:30 - Thanks - วิทยาศาสตร์และเทคโนโลยี
I worked for DEC in the '90's. The local state government invited us and other tech vendors to a "technology update day for local politicians" to showcase new technologies..... in 1994 or 1995 from memory. We decided to showcase the growth of commercial uses of the internet and show the politicians the Web .....and Mosaic (at that stage unknown to any of them). Across from our stand was a competitor showing off Dragon Dictate - via some very attractive blonde ladies with headsets. Needless to say - 95% of the politicians pretended to be enthralled by Dragon Dictate - with barely a politician staying more than 5 minutes with us to find out about the existence of the Internet - typical, but so sad really.....
That's awful, miss out on the most important development the world has seen in a generation for Dragon Dictate of all things. I tried using dictate in the 90s what an incredible waste of my time, it barely functioned for me. Even after I read it what felt like a whole training novel.
There is some unknown force in Physics which makes Dragon Dictate work only at Trade Shows partic in 1990s and early 2000s. A couple of times we saw this at a show and it looked to be an option for a few internal customers for whom regular keyboards not possible. But every time we tried it outside the Trade Show Enabling Field it never worked. My suspicion is that the demonstrators must have spent 100s. (?) of hours perfecting their craft, training the system and its operators.
When it was made to work (relatively recently c.2015) the end user was issued a high-end heavy laptop.
here I am browsing the internet on a DEC inspired CPU with an OS descending from the one the first Web Browser came from
@@john-r-edge
In February 2020 I was able to deliver a 5minute unprepared talk to a friend's laptop. I knew the subject.
It amazed me that it only needed to be cleaned up slightly. Some of that were changes in emphasis.
I had never done that before.
Well it was a cool program. Speech to text was star trek level technology back then
1:10 - Fun fact, I work at CERN, Ive seen the famous Next cube and ive been in Tims office :) Ive also been in a Zoom call with Tim and Vint Cerf
That must be a fascinating place to work. Ed who is in that voxpops section used to work at Rutherford on the MICE project.
I love the fact the early internet was able to be displayed by hand drawn diagrams.
Its amazing that was ever posible.
Even as late as 1992, all of the "interesting" stuff on the internet was kept in a list... the Yanoff List, the precursor to the search engine. IIRC, I used to keep a printed version at hand, and it was only like 4 or 5 pages.
@@buckykattnji still have a copy on my PC, it's very nostalgic to reread!
I imagine a modern diagram would be closer to a fractal at a glance
@@silentantagonist2333 No complete maps of the modern internet exist, several maps have been made and they usually have to be displayed in 3D diagrams but none of them are actually complete as several parts obviously just aren't directly accessible to everyone anymore and it's just so huge that one one has ever been able to compile a complete map.
This is a pretty well researched video. I would like to draw attention to Steve Crocker who led the Network Working Group while a graduate student at UCLA that developed NCP, FTP, TELNET and other application protocols. Crocker introduced the Request for Comments series of notes documenting the work. Jon Postel became the editor of the RFC series and the "Numbers Czar" (eventually the Internet Assigned Numbers Authority). Jon led the effort to create what is now the Internet Corporation for Assigned Names and Numbers but passed away in Sept 1998 just as the organization was being formed. David Clark at MIT became the chief Internet architect when I left ARPA late in 1982 to join MCI, and Bob Kahn, at that time director of the Information Processing Techniques Office (IPTO) of ARPA, continued to manage the effort from the ARPA perspective. Clark served for many years as the chairman of what is now called the Internet Architecture Board but started as the "Internet Configuration Control Board" while I was still at IPTO running the program. There are hundreds of others deserving mention (see authors of all the RFCs). Speaking of which, Bob Braden and Joyce Reynolds were able assistants to Jon and carried on his work after he passed away. Danny Cohen, David Reed and Jon Postel were instrumental in arguing for splitting TCP into TCP/IP in aid of real-time applications such as packet voice, or packet video or real-time tracking.
Well I very much did not expect my video to reach you. So that is a rather pleasant suprise. Thank you for adding the extra information, I would love todo more on the early years of the development of the internet. I don't l know if you would be interesting in being interviewed at some point it would be great to get a insight on what it was like working with all these poeple, as with out someone's first hand experience it's hard to convey that in a video.
@@RetroBytesUK Hi Vint (waves from [ih]). RB, a point you may want to correct: BIND doesn't stand for Berkeley Internet Domain Name (which, if it was, would surely be BIDN ;-)). It actually stands for Berkeley Internet Name Daemon. Good video!
@@niggun996 🤣I spotted that a bit too late as well.
Ummmm - Steve crocker didn't 'develop' the protocols you list -that's laughable. I guarantee Abhay Bhushan or Stephen Carr don't agree with your version of things.
The NWG was the group that developed these various protocols. Crocker let that group. Abhay will certainly agree that Crocker led NWG, but credit for FTP goes to Abhay for sure. Others get credit for specific protocols. I did not say Crocker invented them - but he did lead the group that collectively developed them.@@witness1013
I worked for a couple of companies that dealt with BBN. One of them was a TELCOM provider that made "high speed" modems during the 1980s and the other made TEMPEST terminals and retrofitted other equipment to TEMPEST standards for them.
That was more thorough than I thought would squeeze into 30-odd minutes, and I learned something!
There was alot to get in, and I did not really get into the emergence of transit providers, peering, and commercial ISPs. I think that will all get its own video at some point.
Tremendous video. Anyone interested in learning more about the origins of ARPAnet and the Internet as a whole should certainly read “Where Wizards Stay Up Late” by Hafner and Lyon, which covers all the early BBN stuff in great detail, and explains IMPs and TIPs and so on. There’s a good audiobook version too.
Now that could very much be worth a read/listen.
@@RetroBytesUK Yes, Where Wizards Stay Up Late is an excellent read. Highly recommended.
Added that audio book to my wish list, thanks for the tip!
Norway's NORID had the same strict rules as Australia, you had to have a commercial company single-person business or organisation to get a .no subdomain. And the name had to be related to said operation. Those requirements are reduced now, at least name-wise.
Thanks for mentioning NORSAR; many think the UK were the second country to join ARPANET, and thus, make it international; but (also if you see the map) London connected /through/ NORSAR that had been connected weeks earlier or something.
And I'm glad my ISP gives me a /64 true native IPv6 space so I can have servers ::b00b:1 and ::b00b:2 and ::d3ad:b33f at home BUT public via DNS without NAT and crap. It's annoying when the work ISP only delivers IPv4 but that's not worse than creating a tunnel through a Linode
It was not a big foot print in Norway, but its was very much first, and as you point out the way UCL in the UK connected to Arpanet.
Early 1990s, I attended a meeting in a pharma company meeting in Berlin. When the issue of sharing data files amongst many international affiliates, I suggested Internet standards could be useful, especially with the use of VPNs. There was a longer discussion amongst the group at the top table after which they announced they would only consider using the Internet "when it became stable".
Not that unusual at the time tbh. Many smaller companies had their own ISDN lines connecting offices in different cities and whatnot well into the mid 2000s.
I only wish *I* could meet people like these at one of my local pubs ...
They are a fun group to go drinking with.
Brings back lots of memories of my days at MIT and when the first IMP was delivered as well as networking before.
Quite good though I'm a bit skeptical about V6 adoption because we've gotten so good at workarounds.
Now the emergency ipv4 allocations are more or less exhausted from the likes of ripe, there is a real cost to aquiring more ipv4 address space for ISPs. This will drive up adoption rates amoung ISP (infact it already is). Also mobile providers have move vast chunks of the population on the v6 and nobody really noticed. You are also starting to see some of the larger networks requiring ipv6 for turning up private peering links. It feels like v6 is finally gathering some pace, instead of being the one day in the future protocol.
I was shocked about the stupid idea of removing the 10/8 from the private address space (then obviously not RFC1918 anymore). But I guess it won't come, as I had worked inside a large telecom company / Level 1 to 3 ISP, that uses 10/8 for the intranet and the maintenance of all equipment... And I bet, other large network operators are keeping it the same... In fact, it's the dumbest idea, I had ever heard of... I guess, in the western world, IPv4 will still be around a long time, as the addresses are there for the old ISPs and even IoT does mostly with NAT. It'll be the other parts of the world, which will be way faster on IPv6. But for the management, Marketing will force on additional IPv6 reachability, as, AFAIR, Google would rank them down, if unreachable over IPv6... 😉
Others note that BBN 1822 was the Host/IMP protocol, unchanged for both NCP and TCP/IP. The big flag day was the transition from NCP to TCP/IP. The first three networks of the Internet were Packet Radio Net, Packet Satellite Net and Arpanet. The UCL connection comes in1973 with NCP and TCP/IP in 1982 (ahead of the rest of the Arpanet). In the run up to a major 3 network demo of TCP/IP in November 1977, there were tests of Ethernet to Packet Radio (Xerox PARC and SRI) and Packet Radio to Arpanet using TCP/IP.
Was packet radio net a separately run network ? Its frequently written about as if its part of arpanet. Is that just the author getting mixup about what the arrangements where ?
I must admit I though SatNet connected to arapnet in 77, I take it that date is wrong ?
Thanks for the clarification over the use of 1822, now you say that it now seams obvious that 1822 remained between host and imp as layer 2.
One small error I noticed. ARPAnet IMP 3 was installed at Univ. of California Santa Barbara, not Berkeley. At UCSB, it connected an IBM 360 host computer to the ARPAnet. One of the key features of the ARPAnet was that different types of computers (sometimes from different computer manufacturers and sometimes running different operating systems) could talk to each over a common set of protocols. In the early ARPAnet, this was NCP. Telnet (remote login), FTP (file transfer), and eventually email (after it was invented by Ray Tomlinson at BBN in 1971) all interoperated between different types of computers running different operating systems.
Great video! I live in the Washington, DC, area and there were several IMPs and TIPs around here. If you knew the right people, you could get the dial up phone number for the Mitre TIP or the Pentagon TIP. I used these to access a free account on MIT-AI, a PDP-10, over the ARPANET. Fun times. On a production note, I discourage the use of background music with vocals; they can be very distracting.
What an excellent video! Don't hold your breath for IPv4 to go away anytime soon, I remember playing around on the 6BONE in the late 90s but IPv4 is still the universal standard. 😊Oh by the way, at work we recently purchased a /22 for about 50k.
Its the cost of aquiring v4 space is what will finally drive ISPs to adopt v6. However you are right its still a while away.
I remember videos circa 2008 talking about how the switchover was just a few years away… but of course legacy telcos who had enough v4 space from early on are dragging their feet, making it harder for the small ISPs. I do have a v6 address now for my home network, but I don’t know how much uses it - presumably apps do, from what was said in this video, but I’m not sure about websites.
Neat observation I've made recently: these days, it seems like mobile phones are the most eager adopters of IPv6 addressing, if the network they're connected to offers it. The mDNS definitions broadcast by these devices typically point to their IPv6 addresses, if the WLAN they're on is a multi-stack v4/v6 network, or v6-only of course.
I'm experimenting with setting up an IPv6-only network at home now, seeing as my local ISP supports it and gives me a big block of addresses to play with. It's weird not having to worry about NAT and port-forwarding on routers anymore, which is strangely freeing. But I still miss the 4-decimal-octet addressing system of IPv4 a little, which feels much simpler to understand.
Follow-up to my previous comment: Another thing I kind of miss from IPv4 is the use of dots as the subnet separator character, instead of colons. The decision for IPv6 to go with ":" instead of "." introduces a parsing ambiguity in URLs which necessitates surrounding the IP address in square brackets, so the browser doesn't mistake what you typed for "address:port". I've been getting used to this convention, but it bothers me slightly that this issue could've been entirely avoided had they stuck with periods (or full stops, for those in the UK) as the chosen separator character.
@@xplinux22 You just use brackets [] as such: [::1]:12345/
Yeah. The internet is definitely not as friendly as it was in the mid 90s. It was a great place to explore and make friends.
Yeah. There was a reason all early TCP/IP protocols were open. Nobody ever imagined people abusing it for earnest (e.g. beyond the simple prank). People watched out for each other, but also kept each other on the straight and narrow.
The internet was supposed to make humanity better, but thanks to many things, including social media, it has made humanity, way, way worse.
@@rsr789 Usenet in the 80s and 90s was arguably even wilder than today's social media. I indeed miss when internet was the wild west.
In Brazil the .br domain was also business only, you had to have a CNPJ (the equivalent to EIN maybe) in order to register one. So pretty much the same .au domain model, no random names allowed and as far as I know no obscene as well. This changed more than a decade ago, probably for the same reason as the .au did.
I felt like I was driving through Hoboken watching this video, which I enjoyed. Thanks.
Great video, filled in lots of bits I didn't know. Really enjoy your content.
Thanks Andrew
Thanks for creating this overview. Been online ever since the late 80s, early 90s at Uni and lived through quite a fair bit of it myself. But the very start of it was a lot of new information for me.
Just found your channel - as an old timer who started on IT in the 70’s this video taught me a few things. Fascinating and informative. Thanks.
19:23 JANET had its own network protocol stack, known as the “Coloured Books”, because each protocol was described in a book with a particular-coloured cover. The only one I can remember ever knowing was the e-mail protocol was the “Grey Book”.
They did indeed, JANET X25 and that whole stack is a very interesting rabbit hole I chose not to go into in this video. At some point X25 and JANET are going to get their own video. I think I've touched on X400 mail and X500 directory services in some past videos however.
1:29 “’Ere ... didn’t you kill my brother?”
Loved this video, great to see a more detailed history of networking. Thanks
Great video. I was reluctant to click on it, but glad I did. I first used Telnet in 1992 and I was a BBS enthusiast from around that time too. It's amazing how far things have come. I do however miss the old days when communication between two or more devices was much more civil due to it requiring a certain level of intelligence to ascertain how you could do it ;)
31:20 Actually, there were two different kinds of RRs (“Resource Records”) defined for representing IPv6 addresses in the DNS: there was “A6”, and there was “AAAA”. Wouldn’t you know it, the one with the nice, short, easy-to-say name got deprecated, and the one with the more awkward name became the official way to do it.
I think its because AAAA helps with the transition for v4 to v6 as it works much more like a v4 A record, rather than the A6 being odd split record with n bits of the address coming from the dns server your querying and the rest of the record being the dns server you can request the first n bits from. I don't think I've seen an A6 recording in the wild, I'm assuming bind supports it, its just no one is using it. From a video standpoint I decided just to leave a6 out.
Fabulous. I remember facing articles of incorporation and other identifying info to Internic to register domains as an early web hosting provider like Rackspace in the late 90s
This really filled in some gaps in what was happening that 20-something me couldn’t be bothered to learn back then.
Thank you so much!
LOL 😂😂😂
I looked at the thumbnail and I am thinking to myself: 🤔 "What does the Architect from the Matrix movies have to do with the creation of the Internet?!!"
Then I realized Vint Cerf (a.k.a. The father of Internet) is a doppelganger of Helmut Bakaitis (The Actor who played the Architect in the Matrix movies). 🤣🤣🤣
I just found your channel and I absolutely love it. Please do a video on the WWIV BBS if there’s enough of a story to tell. Thank you for the amazing content!
Lovely and interesting video - quite a few bits I didn't know. Thank you!
Pun intended?
😊 cracking stuff! Really enjoyed that.
Holy crap! I just realized that in the early 2000s I was working in a NOC that was using this "IMP" infrastructure design. Although we were calling them FEP's or Front End Processors. This was all running very legacy communications technologies for the airlines. Shows how much I knew about that stuff at the time, since we were mostly concerned with the message handling protocols, and not the switching technologies.
You might have been using IBM System Network Architecture (SNA). They called IMPs FEPs.
How very Swiss to say, "We're waiting for a standard."
I watched this video over IPv6. I'd say 60% of my traffic is IPv6, with the rest having only IPv4 DNS records.
I was running a 6 in 4 tunnel with Hurricane for my IPv6 connectivity, but Hurricane shut that service down. So I wont get IPv6 back again until I move provider.
Many big organizations were waiting for the ISO-OSI protocol stack to take over. TCP/IP was seen by many as very much an interim standard, put together by a bunch of academic neckbeards, that would be supplanted by “proper” specs drawn up by “official” standards bodies like ISO. But OSI became that ultimate bureaucratic horror-a standard created by a committee-only much worse: an unwieldy bunch of complicated, interlocking standards designed by hordes of different committees.
In the end, only fragments of OSI still remain-who still uses X.400 mail, or remembers what TP4 is for? The most visible piece is probably the OSI seven-layer “reference model” for how the protocol stack is put together. But there is another (partial) piece: the X.500 Directory Access Protocol was subsetted to produce the Lightweight Directory Access Protocol or LDAP, and that is pretty much the standard for network directory services today.
Always good to see the history of development, especially when so well presented! 😄
Thanks Ponder. Its a particularly interesting history, that given how we all use the internet so few poeple know.
ARPA net connected military with itself, it's researchers at Universities and it's contractors. Thus at Northrop we used ARPA net to do commercial activities with the Military. Before that at UCLA we used ARPA Net to do research, some of this "research" were games. NSF net was developed for non military research so that you didn't have to get a military contract to use ARPA net.
University researchers who got contracts from the military and grants from NSF got frustrated that they had to switch between the separate networks, so an internet connection was made. In the beginning of this connection researchers referred to there ability to work across the ARPA and NSF nets as "the internet "
I love these. Thank you.
I caught the Gopher reference. I used it in the early 90'S at Univ of FL, on a Vax, I was fascinated with the amount of information available, and the bulletin boards.
Great vid RB, thank you!
Minor point, thanks for being nice to people like Elizabeth, means a lot to see a TH-camr supportive in multiple videos of our historical figures.
Second, when I was studying ipv6 they mentioned ipv6 bridges which I think will help a lot of the issues that we have with ipv4 adoption.
My university, UMaine, was able to use PDP11s, and later some VAX machines in the 80s to hook up to BITNET. BITNET wasnt mentioned, but it was the most popular unaffiliated online network in the 80s. You had email, usenet, ftp, chat groups, message boards, etc. And online games.
I think I will do a video on BITNET at some point, but I need to gather a bit more background on it.
Once apon a time in the mid 1990s I was working in the research laboratories of BT in Martlesham Heath near Ipswich, and Vint Cerf was visiting (he worked for MCI, which was 20% owned by BT then).
I remember him giving a really great presentation about the Internet, but I unfortunately can not remember anything about it, other than it was Vint on stage!
Excellent video, I use to teach most of this to business and end users in the 90s to help them understand why they should invest in it at the time. Being a yank I usually gave credit to ARPA ;) And I agree, I got online in 88 and, through on old PDP, and it was a much different and nicer place most of the time.
Back in the early 1980's, I was at MIT, but not working in computer science. Even for me, BBN was a name that I remember, as they were huge at Cambridge. Back then, we had various machines (PDP, and starting to get some PCs), but nothing was networked. I used to talk to other students who were using various other networks (bitnet, etc), to talk to stuff around the world - it wasn't until several years later that I was working at a place where networking was ubiquitous.
I remember dealing with those damned hosts files too. And then you get the phone calls from someone trying to send you an email but unable to because their hosts file was out of date.
DNS made life so much easier.
At about 24:00, I was desperately hoping that the department spun up by the Aussies to handle Kevin's job was named "KEVIN".
18:50 A big change which happened sometime mid-1980s was the opening up of the Internet to commercial use. There were big companies (particularly the computer ones) on the Internet before then, but I think they were never allowed to connect directly to each other, they always had to go through some research institute or university or such. Once the restrictions on who was allowed on were dropped, that’s when the modern era of the Internet really began.
They where always allowed to take private interconnects between each other, but hardly ever happened for example ICL had direct connection to the DSS, was well as a link to University (I don't remember which one), however as you said the allowing of commercial use drove much wider adoption.
Minor correction, there is a chunk of private IP space in IPv6. The ULA block fd00::/8(or more technically fc00::/7) specified by RFC4193. That being said NAT doesn't exist so the original private use case is the only use case for ULA addressing now. Internet access requires devices to have a global IP so if you want to use ULAs and want internet you have to assign both IP types, a ULA and a global one. Some people argue ULA space doesn't need to exist...that's...largely true but it does have some legitimate use cases mostly around local numbering independence when your ISP does what they're not supposed to do and assigns you a dynamic prefix. You can use the ISP provided prefix for internet access exclusively and then use a ULA prefix for your local networking needs since IPv6 fully supports multiple IP addresses on one interface. Personally I'm of the belief that most of the time you don't need ULAs...only certain select instances does it make sense, link-local and global addressing is suitable for 99% of use cases.
Vint Cerf and Bob Kahn are up there with Turing and Shannon. They architected a system that is as important historically as the industrial revolution.
Nice video - thanks for sharing B)
For the last 12 years I've had a birthday reminder for May 22. I'm only slightly older than Ethernet.
Acutally there're a private IPv6 adress range called ULA, Unique local address and is the same as in IPv4, despite not really necessary. Theses adresses begin with fc or fd. For me they're quite handy because my v6 prefix might change, but the ULA prefix can be fixed.
16:48 As if it's not bad enough handing out Class A networks to big companies like candy, you can't even use all of them! I get setting aside some addresses for multicast, but the real offense is Class E space. We're all sitting around saying "despair and lament, IPv4 is exhausted" and meanwhile 240/4, one sixteenth of the entire range, is pristine. What a shame.
I set up the first Internet and LAN at my University back in 1987 or 1988, something. Set up the first email server and domain back then.
Internet started first, around 1980.. Webb started back in 1991, and was built on Internet. But Gopher was a couple of years earlier then the Webb.
The French CYCLADES project also had a major influence on the development of TCP/IP.
Huh, wow. 18:01 There was that graph. I had public IP address for my home server back in -98. Looks like routing requirements have grown a bit after that ;)
By the way, the problem with IPv4 is NOT the limited number of addresses but how generously they were allocated back in the day. Even my university in Turkey had a Class B IP space of its own. They were using real IPs not just for their servers but also for all the workstations, students connecting from dorms and what have you. IPv6 can easily be exhausted if the same mistakes are repeated. The adoption however has been going slower, thanks to technologies like NAT that made it possible to connect entire networks with just one or few hosts connected to the ISP.
Most of those overly large allocations got clawed back by their regional registrars quiet a few years back now. A fair few companies and institutions where originally given class B or worse subnets back in the early days. IPv6's address space is huge however, to the point where is has been estimated we would allocate 1 v6 address for every atom in the universe and still not have exhausted the address space.
Excellent 👍
Gopher and Archie 🤣🤣🤣
Good video, thanks.
Because of your video I discovered my school (Purdue University) has been on the internet since 1975! We have also held on to the 128.10.X.X IP range for upwards of 40 years!
Readers: Vint Cerf commented, but he's in the video. Steve Blumenthal was an early important manager of ARPAnet and following and Bob Frankston had something to do with electronic spreadsheets. :-)
5:42 I think the IMP was more like a modern network interface card. In those days, you couldn’t pack that much electronics on a single expansion card, so it had to sit outside the main computer cabinet in its own box.
From talking to people involved with the IMPs back then, an IMP would have one or more 1822 interfaces in it, and the hosts would have an 1822 interface. Apparently we would not be far off the mark if we thought of them as serial ports. There was then also 1 or more modems to link them to the other IMPs, as initially they just used regular phone lines for the interconnection. Its just there was no dialling as the line where permanently connected. What I could not find out was the details of the modulation scheme employed, so I don't know if it stayed with in the normal frequency restrictions for a phone line.
I did explain the transition poorly for the change over to IP. In that 1822 did not go away it just became the layer 2 protocol between IMP and host. Or at least it did for the hosts still connected to an IMP.
@@RetroBytesUK Simple answer: not "regular" phone lines, no modulation involved... Having been the Network Manager at MIT's LCS when the ARPAnet was being decommissioned (all the ARPAnet stuff at MIT was in our building), I got to know a lot about the IMP/IMP links. There were three "modems" in a large (approx 2 yard/meter cube) that had four roll out frames (one empty) the "modems" were owned by the phone company and part of the leased line digital link service, two labelled 56Kbps and one 50Kbps. They weren't actually modems since they were digital on both sides, so no modulated voice frequency signal. There were 3 IMPs at MIT at the time with hardwired connections between them and the three modems were for links to remote IMPs. The digital link service was using the same base connection that the phone company used internally for carrying digitized calls (called a DS0), which was originally a 50Kbps signal and later a 56Kbps signal, I never found out whether the modem tagged 50 was actually still running that speed or whether it had just been upgraded without relabeling.
Most untold story: everybody has it, nobody knows where it comes from
Good stuff.
5:04 Another point about packet networks is that delivery was on a “best-effort” basis. That means, there was no guarantee that any packet would actually arrive. Instead, you built higher-level protocols on top of your packet protocol that provided their own reliability guarantees. Or timeliness guarantees, if that was more important than reliable delivery (the tradeoff is described as “quality of service” or QoS).
This horrified the telephone-engineer types. Remember, they dealt purely with circuit-switching, and circuits were always reliable-stuff you put in one end was guaranteed to make its way to the other end without getting lost. They couldn’t understand how an unreliable packet service could ever be of any use.
Today, even the phone networks are adopting VoIP, which runs over-you guessed it-the unreliable packet service offered by the Internet.
Very true, its amazing how many VoIP engineers forget the potentially lossy nature of a network an get rather upset when the thing that is allowed to happen (a packet not arriving) happens.
Great video.
...*administered
/runs away!
I knew quite a lot of the story already, but this was a great video nonetheless. And now I know about Kevin! Happy days!
I feel like everyone needs to know about Kevin.
I remember in 80's early 90's a lot of companies had their own text-only bulletin boards. I had friends who ran two major companies' boards. very basic stuff but it was very interesting at the time.
I have a big soft spot for BBSs
Learning about obscure IT history facts while listening to Mafia music feels like bringing two incompatible friends along with me for drinks. So no, I can't complain
Mafia music? You mean Django Reinhardt? Cuz he was a Belgian jazz musician / compaser who died in 1953, and had nothing whatsoever to do with 'the mafia'.
@@rsr789 Yes! I do mean Django Reinhardt, whose music I came to know and love via Mafia, the 2002 PC game, whose OST had quite a few of his pieces. Rest his soul, I don't imply that he was member of a crime ring. I think the channel owner got the reference though
What's the background music? It's a very interesting choice. The old-timey sound is quite a contrast to the subject matter!
25:21 That's the thing though: people on Twitter don't lose their cool and SHOUT IN ALL CAPS in that way that people were (supposedly) prone to do on Usenet in the old days. (Ofc Usenet was originally invented not on ARPANET but on the UUCP network, but that's another story.) I think that people on the Internet have figured out that obviously losing your cool and expressing unbridled anger will tend to get you banned for bad behaviour, and even worse will tend to get you mocked. Raging is an expression of emotional vulnerability: it's as good as admitting that the other guy has managed to get to you. But if your usual mode is sarcasm then you can always conceal anger or upset behind the mask of sneering, effortless disdain.
cool stuff
There is a YTer I love called The Tim Traveler. He did a video about the *grandfathers* of the internet. It was an supreme video and would pair well with this.
I'll have to watch that, I really like his videos. Although most the ones I've seen are about obsucre board issues.
Here in the US most histories credit Paul Baran of RAND with inventing the concept of packet switching. We also take full credit for digital computing.
Tis similar to the RSA Public Key cryptography that enabled e-commerce, where Cocks invented the algorithm and mechanism, in secret at GCHQ a few years before the public re-invention of the mechanism and naming in the USofA.
Baran similarly was the first to start working on the concept of packet switching, in secret for the US military, but Davies independently re-invented the concept, named, academically published, presented, promoted, and built a practical network, a few years later.
I remember BBN from the history of Lisp. They made BBN Lisp.
First time on internet was on the A1200. SCSI connected via PCMCIA, and being I had 3.0 Workbench and ROMs, the system would crash if I connected anything into the PCMCIA socket whilst having more than 4mb fast RAM due to a bug in the OS ROM that has a conflict between upper RAM(any fast RAM above 4.5mb) and PCMCIA on the system bus. So yeah, web browsing could be slow with plenty of thrashing(I had to use a third party app to get a page file), even at 640x288.
I also used an amiga one of the first times I used the Internet at home. My A500 was not up to running a Web browser, but I could use ftp, email, and a newsnet client.
@@RetroBytesUK This was pretty much, along with SWOS, Worms and Doom, my A1200s last stand.
By this point I had an '030 and 68882, but even then I got better performance from Doom at 640x480i(completely unplayable, for the record, maybe 2 fps) than that Netscape knockoff I got off an Amiga Format cover CD, but can't remember the name of. Even just trying to scroll the page caused the system to struggle.
@@fattomandeibu I did use A-Mosaic and iBrowse on an A4000 we used in work to prep the disks we gave to Amiga customers of our ISP. A-Mosaic was incredibly behind the times then, and iBrowse was far more modern but still a long way behind PC browsers for Win3.1/95 and Linux.
As an ISP we support a broad range of platforms (this was in 96), we supported DOS/Win3.1, NT workstation, Mac 68k and PowerPC, Linux, AmgiaOS 68k & PPC, RiscOS and BeOS. We also did generic Unix support, but Linux we supported distros with setup instructions (Redhat, Debian, and Slackware).
8:31 what were the physical channels between the universities?
They would have been things called “leased lines”, I’d say.
The biggest missing part in this video: no direction mention of IETF/RFCs
for those not realizing just how WOW! the internet really is, may i direct your attention to the song 'Welcome to the internet' by BoBurnham
25:24 this document has at least 21 pages. If the netiquettes where generally that long I bet almost nobody read them. Probably their main purpose was legal.
I'm pretty sure some PARC staff were also unofficially involved in the planning of TCP/IP, as they'd already learnt a lot from their own setup, couldn't talk about it to competitors, but would hint like hell😂 I think PUP & UDP may have some similarities. XNS was advanced af.
yes, bob metcalfe, john shoch came to my internet seminars at stanford' yogen dalal was a graduate student who worked on internet while at stanford and then joined xerox.
14:39 and thus we have the first simps on the Internet.
Would it be possible to start normalizing the audio for your videos across the full timeline moving forward? Absolutely love your work, but I’m a truck driver and often listen to spoken YT pieces, and when you use the white noise transitions early on in the video, they were much louder than the spoken portions, and nearly caused me to jump out of my seat :P
JANET was an "X.25" network--I couldn't figure out what you had said, and the auto-captions didn't get it either. 19:20
Internet? Never heard of it.
If only it had taken off.
@@UncleKennybobs With the blinking light on top?
9:07 “MAXC” (upper left) ... that was the PDP-10 clone that Xerox built, rather than buy an actual PDP-10 from DEC.
In those days, companies didn’t claim “intellectual property” and sue you ...
The Guy from the matrix help create the internet 13:15 lol.😮💨
😅
Gopher is awesome!!!
Background music is slightly too loud. For a non-native English speaker it's hard to listen to.
Excellent story. I'm a fan.
Thanks for making this. Though, I think given your introduction critiquing general lack of understanding of how the internet was formed must have felt to others as well as, coming over, a little self confident, shall we say.
Not a single mention of DARPA or the fact that it was the millitary considering it's vunerablity to a tactical Nuclear strike (targetting its millitary communication systems, parts of which relied on the then ARPA network) that comissioned the project that became TCP.
It's a core peice of the founding of the modern internet (and this is Modern circa 1980)! Critical infact. Where it not for the driver of the millitary seeking a protocol that would not be subject to disruption if one or more hub communication centers where to be eliminated by enemy attack and the goal for all communication to be automatically rerouted so as not to be vunerable to single strike scenarios (e.g. highly redundant and distributed) then modern TCP/IP simply would not exist (or at least would not be as it is).
Incidently, I was at University when IPv6 was in the process of standarization. Whilst IPv6 was original invented to solve the problem of the shortage of IP addresses, during standardization it became politicised as governments saw IPv6 as the opportunity to retake control of the internet which was largely outside their governance. I recall that IPv6 had, at that time an optional security module which gave increased traceability and tracking designed (by the professors) as an add on to assist with diagnosing for Network Administrators. The US and UK government insisted that IPv6 had to include the module and they saw how IPv4 was, to their thinking, to open to the public (or, if you prefer, as they stated 'to abuse'). So they passed laws to force anyone implementing IPv6 to have to include the security and tracing module. The reason IPv6 has not taken off, is because people who understood technology just simply don't want what it became post politicization; However, government laws are increasingly forcing the issue as much as possible (and its rumored to be the reason that IPv6 is not disableable in modern versions of Windows, and why IPv6 is fundimentally tied to OS system components in unnecessary ways, to make its real / true removal very difficult and bound to break basic functionality not associated with its use). This is also why IPv6 is the required protocol in China.
actually you are referring to work by Paul Baran at RAND on a resilient mesh using store/forward methods. He never actually got to build it. The Arpanet was intended for resource sharing among the research institutions supported by DARPA's Information Processing Techniques Office. Internet was a significant extension of the original Arpanet concept to include multiple networks - but it was not really designed against survival after a nuclear exchange.
Those were the days.
I remember using Gopher to get http server addresses. Thank God for Altavista.
16:36 The classes do not nest. They are entirely separate address blocks, not a hierarchy.
I really don't think that's true, I know for certain ICL's class A and sub classes where hierarchical, I can also be fairly certain that's how it worked for BBN too (as that's what some of their staff have said in talks), JANET's staff have also said that is how their first implementation of IP over X25
worked.
@@RetroBytesUK I was around at that time, so I remember how they worked. Classes were assigned to particular address ranges, with no overlap. You can look up the details in the usual places.
11:44 What’s new? Look at all the CDC machines joining in. And one DG Eclipse, and one or two other new brands. Also the “C.mmp” experimental parallel supercomputer -- seems the ILLIAC IV super from the previous diagram is gone.
I think “XGP” stands for “Xerox Graphics Processor”.
I was just trying to show ArpaNet's growth more than anything else with that visual, there where no particular changes I wanted to draw peoples attention to that I had not already brought up.
aww, poor snorkers! 😢
He never can get Stu to serve him. I have a whole vidoe where thats the joke the whole way through.
As an Australian I will always lament that our country code wasn't .Oz - Austria should've got .Au
On paper IPv6 sounds like it will endure for a long time. In practice my ISP on my small VDSL2+ connection in Australia give me 256 /56 subnets, which is half of the bits. I have the address space to become the ISP for some intergalactic civilisation. Anyway, when things are done in practice like this I would not be surprised by address space exhaustion issues for IPv6 by the end of this century.
What is 256 / 56 subnets?
TBH, Kevin was probably better at managing .AU back in the day than AuDA are today...
Also, slight nitpick 32-bit -> 64-bit doesn't DOUBLE the address space, you're off by a few orders of magnitude ;)
And whilst IPv6 has been *supported* by all the OSes since the late 90's, most of the implementations were garbage until about 5-10 years ago, bit of a chicken-and-egg problem.
Doubles address space consumed in memory per address. Otherwise, I agree -- what's a few orders of magnitude among friends? Me? I'm an old IMP configurer.
Double the address space of 32 bit is 33 bit !!!
You have not mentioned Man-Bear-Pig, the inventor of the internet!
Why haven't you mentioned the monstrocity that is NAT?
We started our own ISP in probably 1993. Eventually had our own AS. Good times. Unfortunately my partner screwed me over and kept most of the money we made for himself.
Its a shame it ended that way for you. As someone who also started an ISP at that time, I completely agree it was a great time to get involved.
@@RetroBytesUK Thanks. Should I ever get into the situation of starting a potentially multi-multi-million company I‘ll try to prepare better :) But oh well, at least I had a leased line at home in the 90ies, that‘s something, isn‘t it :)
I really don't see why 32 bits isn't still enough. For god's sake, that's four IP addresses for EVERY PERSON ON THE PLANET. If that's not enough, then we're doing something wrong, and there was a better way to fix it than expanding the address space.
We do have vastly more devices on the internet than will fit into 32bits already thanks to NAT. We are now basically out of public address space for ISPs to give addresses to just the routers. especially if its not a long established ISP. Part of it is for each public subnet we loose 2 IPs one for the router providing service to the other addresses in that subnet, the other for the broadcast address. So there is some inefficiencies. However if you think about how many devices in your house have an IP stack, and think of that replicated across every house its far more than 2 per person. Even my parents have about 20 devices in their house with IP addresses.
Was that Ron from Sparks?