0:00:00.000,0:00:04.079
So, welcome back. Now the purpose, of[br]course, of all this stuff that you had in

0:00:04.079,0:00:09.091
your office, and all that banging, and all[br]the sqweeee squealing noise where we're

0:00:09.091,0:00:15.046
sending the data across the across the[br]phone using sound. All the purpose was the

0:00:15.046,0:00:21.010
fact that computation was rare and[br]extremely valuable. And for scientists who

0:00:21.010,0:00:27.005
were trying to solve research problems,[br]access to computers was essential. And you

0:00:27.005,0:00:33.021
couldn't all sit, sort of in a little ring[br]around the computer. It's so we would have

0:00:33.021,0:00:38.089
phones in our offices and we would, you[br]know, work in our office, just like I'm

0:00:38.089,0:00:45.024
working in my office right now. But there[br]wasn't enough computation [inaudible], any

0:00:45.024,0:00:51.058
work in my office so I had to connect to[br]something outside. And so this was the way

0:00:51.058,0:00:57.078
of science. And, it, it, it was the fact[br]that comp-, computing was rare, and access

0:00:57.078,0:01:03.066
to computing was a critical enabler of[br]scientific research. Now, I'd mentioned

0:01:03.066,0:01:09.075
earlier, data transfer with leased lines.[br]And so, while it. You tended to interact

0:01:09.075,0:01:14.045
with the computer that was rather local,[br]geographically to you because you could

0:01:14.045,0:01:19.009
have this permanent dial up connection all[br]day long without paying a permanent

0:01:19.009,0:01:23.096
charge. If you were a bank, or you had[br]some really critical need you would lease

0:01:23.096,0:01:28.656
some line from the phone company 24 hours[br]a day, seven days a week so you could send

0:01:28.656,0:01:33.044
data across that anytime that you wanted.[br]No dialing, it's always connected and,

0:01:33.044,0:01:38.618
after while, I mean you can send data, we[br]academics wanted to communicate with each

0:01:38.618,0:01:43.938
other. It would be nice to be able to use[br]each other's computers, but we tended to

0:01:43.938,0:01:47.898
have too much, but sometimes we don't want[br]to send a file, or some email, or

0:01:47.898,0:01:52.623
something like that. And so this led to[br]the invention and the creation of store

0:01:52.623,0:01:57.772
and forward networking. And how this would[br]work is we would sort of, you would sort

0:01:57.772,0:02:03.796
of have some thing and you would use a[br]modem to do all your dialing, like that. I

0:02:03.796,0:02:08.957
mean maybe they had paper and, and didn't.[br]That looks a little too advanced, but you

0:02:08.957,0:02:13.175
would have some geographically local[br]computer that was your, sort of the compu

0:02:13.175,0:02:17.237
ter that did most of your work. You didn't[br]have a computer in your office, you just

0:02:17.237,0:02:21.242
had a connection to the single campus[br]computer or on a few campus computers. And

0:02:21.242,0:02:25.753
then what universities would do is they[br]would lease a line. And then we could

0:02:25.753,0:02:31.183
send, write a mail program that would run[br]on this computer and then would send mail

0:02:31.183,0:02:36.320
and then everyone else would read it. And[br]what happened was is we sort of started

0:02:36.320,0:02:41.835
stringing them together in these snakelike[br]structures and so, we could share this.

0:02:41.835,0:02:47.545
And so let me just show you kind of how[br]the store and forward networking works. So

0:02:47.545,0:02:52.316
somebody sends a mail message in. Now[br]let's say, let's say we are this bottom

0:02:52.316,0:02:56.982
person down here, okay. And so someone[br]else has sent a mail message in, they're

0:02:56.982,0:03:01.460
sitting in there. And now the next person[br]sends a mail message in and now ten

0:03:01.460,0:03:05.794
seconds later, you send a mail message in.[br]Those mail messengers are sitting in a

0:03:05.794,0:03:10.665
cue. They're waiting just like waiting in[br]line at the bus stop or waiting in line in

0:03:10.665,0:03:15.269
a train station. Waiting in line for a cup[br]of coffee at Starbucks. They're waiting in

0:03:15.269,0:03:19.498
line, and what would happen is the[br]computer that was our local computer would

0:03:19.498,0:03:24.498
then start sending that data across the[br]line. Okay, and slowly but surely it would

0:03:24.498,0:03:29.686
take awhile. And everyone else had to sit[br]and wait. Your poor message is last in

0:03:29.686,0:03:34.199
line so you have to wait. So finally this[br]message gets across the closest link. And

0:03:34.199,0:03:38.878
then the, the next message starts being[br]sent and you have to wait for that message

0:03:38.878,0:03:43.495
and wait and wait and wait and wait and[br]wait. Wait and wait and wait! Hey wait.

0:03:43.495,0:03:48.807
Okay. Now its finally your turn. So your[br]message finally gets to use the one

0:03:48.807,0:03:53.162
connected line. So they, they are sought[br]of stand in line until your turn

0:03:53.162,0:03:57.736
[inaudible] runs acrossed. And they all,[br]these messages aren't destined for just

0:03:57.736,0:04:02.932
one computer away, then they got to go[br]through the whole thing again, move across

0:04:02.932,0:04:07.748
the next link until you know eventually[br]you move across one link over here and

0:04:07.748,0:04:12.396
then go another, and then finally talks to[br]the people who get their email. So its a

0:04:12.396,0:04:17.415
sought of dedicat ed line and you had to[br]stand in line to get your chance. And the

0:04:17.415,0:04:23.138
key thing here is each of these lease[br]lines has a fixed cost 24 hours a day

0:04:23.138,0:04:28.940
seven days a week, and it's very dependent[br]on the distance, so we saw a weird

0:04:28.940,0:04:33.464
phenomena. ≫> And that is. ≫>[br]If we could add hops, it would slow our

0:04:33.464,0:04:38.491
message down, but it would reduce our cost[br]greatly. And so let's just say we have

0:04:38.491,0:04:43.985
Michigan State University, which is where[br]I got all my degrees from. University of

0:04:43.985,0:04:48.944
Michigan here in Ann Arbor, which is where[br]I work. And let's say, you know, we're

0:04:48.944,0:04:54.081
connecting to the rest of the world, and[br]we're going through Cleveland, where Case

0:04:54.081,0:04:59.566
Western Reserve is. Case Western Reserve[br]was the early innovator in, in networking,

0:04:59.566,0:05:04.019
and so we have two leased lines with a[br]certain distance, right? One from East

0:05:04.019,0:05:08.646
Lansing to Ann Arbor and one from Ann[br]Arbor to Cleveland and so we're sharing

0:05:08.646,0:05:14.112
the cost of these lines between three[br]schools and we can all kind of connect to

0:05:14.112,0:05:19.427
the rest of the internet, all connect to[br]the rest of the internet out here and, and

0:05:19.427,0:05:24.029
we just, some of us have, are farther[br]away, and so we take longer. The folks in

0:05:24.029,0:05:28.921
Cleveland are closer. Like all the rest of[br]the connection to like the East Coast and

0:05:28.921,0:05:34.955
the West Coast come through, say like here[br]Cleveland, but if we can simply convince

0:05:34.955,0:05:43.118
somebody in between us like say Toledo to[br]add a connection. Now of course. Of

0:05:43.118,0:05:51.095
course, this, [inaudible] Give me green.[br]There we go.'Course this line probably

0:05:51.095,0:05:56.312
goes, probably went around when we just[br]went straight to Cleveland, here. But

0:05:56.312,0:06:01.815
basically if we can convince Toledo to[br]sort of put in their computer and hold

0:06:01.815,0:06:06.892
onto our messages for a while, we could[br]send now one hop, two hops, three hops.

0:06:06.892,0:06:12.171
But the cost now is not that different,[br]because the original long line between Ann

0:06:12.171,0:06:16.645
Arbor and Cleveland was distance[br]sensitive. And, so, you can think of this

0:06:16.645,0:06:21.800
as, you can get this almost for free. And[br]now we have a whole additional university.

0:06:21.800,0:06:26.711
Both to send stuff to, and they can send[br]to the whole world as well. And so this

0:06:26.711,0:06:31.872
motivation to effectively take the same c[br]ost, and now basically we're taking this

0:06:31.872,0:06:36.784
cost, and dividing it by four schools. And[br]if you start thinking about it, it doesn't

0:06:36.784,0:06:40.857
take long to say, "You know what, let's[br]put one here, one here, one here, one

0:06:40.857,0:06:45.588
here, and one here." Because the cost of[br]the phone company isn't that different.

0:06:45.588,0:06:50.177
You can think of each of these as adding[br]some delay to your message. You know, and

0:06:50.177,0:06:54.868
given the fact that each of these[br]represents an outbound queue of messages

0:06:54.868,0:07:00.478
that are waiting to be sent, there's some[br]delay. There's some cost adding this, but

0:07:00.478,0:07:05.929
It's so much cheaper. So our faculty have[br]to wait another twenty minutes to get

0:07:05.929,0:07:11.126
their mail through if we can bring that[br]many more universities on. And so this

0:07:11.126,0:07:15.881
just works out. There's this sorta[br]motivation that if you can find an

0:07:15.881,0:07:19.992
intermediate person, geographically[br]intermediate school or university or

0:07:19.992,0:07:25.133
company, and you can add them in, you can[br]replace one long link with two short

0:07:25.133,0:07:30.801
links. And this led to long chains of[br]mail. And so from the mid 70s to the late

0:07:30.801,0:07:37.095
80s most academics were communicating[br]through a network that was like this. It

0:07:37.095,0:07:43.450
typically was email and I recall when I[br]first started to use national email. It

0:07:43.450,0:07:49.510
took a long time for mail to go back and[br]forth but it was actually quite magical I

0:07:49.510,0:07:54.102
mean who cares if it took an hour. Now we[br]expect it in three seconds. We send an

0:07:54.102,0:07:58.243
email and hit the refresh buttons, hurry[br]up, hurry up. You know, it could be hours,

0:07:58.243,0:08:02.293
it could be days if you were going far[br]enough and your message was long enough

0:08:02.293,0:08:06.276
and you end up behind too many queues. And[br]so you had this one computer locally and

0:08:06.276,0:08:09.856
every once in a while you'd do most of[br]your communication computation locally.

0:08:09.856,0:08:14.303
And every once in a while you would fire a[br]note off and that would kind of fight its

0:08:14.303,0:08:18.571
way through all those successive[br]connections. This is sort of the life in

0:08:18.571,0:08:24.093
the early 1980s. One of the, most widely[br]distributed networks of this kind was a

0:08:24.093,0:08:29.024
thing called Bitnet. And Princeton was[br]kinda the hub of this and these tendrils

0:08:29.024,0:08:33.657
of connections ran out from Princeton. And[br]by connecting to a, a network with lots of

0:08:33.657,0:08:38.479
oth er folks, then you had more people to[br]talk to. And the more people that you,

0:08:38.479,0:08:43.540
that were connected the cheaper that it[br]was for everybody. So it was a pretty, it

0:08:43.540,0:08:48.407
was the perfect kinda thing that caused[br]people and com-, universities to want to

0:08:48.407,0:08:52.841
work together, because together their[br]shared cost was much, much lower than to,

0:08:52.841,0:08:58.002
to provide this uniform connectivity and[br]email. So at the same time, during that

0:08:58.002,0:09:04.275
same period, where most of us were using[br]store and forward network, with our one on

0:09:04.275,0:09:09.557
campus computer, a bunch of computer[br]scientists were funded by DARPA. The

0:09:09.557,0:09:14.482
Defense Advanced Research Projects[br]Administration, to imagine a different

0:09:14.482,0:09:20.559
kind of network. And the idea was direct[br]connections are expensive. The long trails

0:09:20.559,0:09:26.166
of store and forward networks, they're[br]very slow, and if you had a giant message

0:09:26.166,0:09:32.062
that you got behind, then what, how do you[br]get past that. It could clog the system up

0:09:32.062,0:09:37.318
for, for hours, if not days. And, and how[br]do you keep from failures breaking the

0:09:37.318,0:09:42.443
entire system? If you think about a store[br]and forward network, one computer going

0:09:42.443,0:09:47.515
down would cause data to back up on both[br]sides of that computer until it's done.

0:09:47.515,0:09:52.391
And so, you don't really wanna have one[br]outage and, and how if we have sort of

0:09:52.391,0:09:57.460
instead, instead of just a few messages,[br]what if we just wanted all the messages to

0:09:57.460,0:10:03.214
go simultaneously, so that there's more of[br]a fair allocation of the network, rather

0:10:03.214,0:10:09.124
than whoever gets there first gets it all[br]until they're done with it. And, and so

0:10:09.124,0:10:14.951
Darpa wanted to solve the problem of[br]outages. You know, many will say that it

0:10:14.951,0:10:21.549
had to do with, battlefield conditions,[br]which is probably true They expected that

0:10:21.549,0:10:27.784
various connections would go out in, in,[br]in dynamic situations. Maybe it was that

0:10:27.784,0:10:34.027
stuff was moving. But also how to be more[br]efficient. And so, in effect, you can kind

0:10:34.027,0:10:40.393
of think of this as all a game, where the[br]phone companies own the wire. So everyone,

0:10:40.393,0:10:45.558
even government, even military has to[br]lease the wire from the phone companies.

0:10:45.558,0:10:51.274
And so everyone is like doing research to[br]figure out or creating systems to figure

0:10:51.274,0:10:56.625
how not to pay t he phone company so much[br]money, okay. So these research networks,

0:10:56.625,0:11:02.799
and so if we look for example at this one[br]down here by 1972 they had this network. I

0:11:02.799,0:11:08.396
have my, I, yes I got a caller. So they[br]have this network by 1972 and it's got,

0:11:08.396,0:11:14.076
like some [inaudible] right around twelve,[br]fourteen, fifteen hosts in it, and it goes

0:11:14.076,0:11:19.028
cross-country. Now, now, the, the key[br]about this is in 1972 to have leased lines

0:11:19.028,0:11:24.428
that were up 24 hours a day, seven days a[br]week, all the way across the country? Very

0:11:24.428,0:11:29.258
expensive. But hey, it's a government[br]project, and the government says this is

0:11:29.258,0:11:34.169
important so we're gonna spend the money[br]because, so we're imagining battlefield

0:11:34.169,0:11:39.379
communications of the future and our own[br]ability to do computations so they could

0:11:39.379,0:11:44.098
have comp-, computational equipment all[br]over the place. So this was very

0:11:44.098,0:11:49.365
expensive, but research dollars were being[br]flooded into it, because the q, they were

0:11:49.365,0:11:54.291
solving a research question. If you just[br]think about this as a network, it was not

0:11:54.291,0:11:58.636
all [laugh], it wasn't sorta like, it was[br]so costly that the average person wouldn't

0:11:58.636,0:12:03.343
like, pay $fifteen a month to use it. It[br]would just be that costly. But it's okay.

0:12:03.343,0:12:08.455
Now if you look at this, you see that[br]across the United States, there was always

0:12:08.455,0:12:12.436
at least one connection. They had three[br]cross country links with totally

0:12:12.436,0:12:17.381
independent cross country links, with the[br]ideas that you could take one of these

0:12:17.381,0:12:21.894
things out, and you could still be[br]functioning. So they, they were able to

0:12:21.894,0:12:26.797
research all these things right, as well[br]as the efficiency problem, which they

0:12:26.797,0:12:32.501
solved using packet switching. So, by the[br]mid 70's there was quite a few folks on

0:12:32.501,0:12:37.751
this. And for a group of people they just[br]started using it in production. It was

0:12:37.751,0:12:42.649
pretty cool, right? If you were, if you[br]were one of these universities or

0:12:42.649,0:12:47.801
companies, you had a pretty cool,[br]futuristic world. You could, you could

0:12:47.801,0:12:53.300
send email and get an answer back in two[br]minutes, or a minute, or 30 seconds, even.

0:12:53.300,0:12:58.520
And so it was kind of this futuristic[br]world that was heavily subsidized by the

0:12:58.520,0:13:04.564
government in the name of researching. And[br]so there are two essential things that

0:13:04.564,0:13:10.747
really came out of this research. And one[br]is the notion of what was called Packet

0:13:10.747,0:13:16.222
Switching. Packet Switching basically[br]eliminates the problem where once the

0:13:16.222,0:13:22.066
message starts using that leased line[br]wire, you have to wait till they're done.

0:13:22.066,0:13:27.933
As, as I showed in that in that store and[br]forward. What you want is to be able to

0:13:27.933,0:13:33.101
send little pieces. Break the messages up[br]into little pieces, and then they, they

0:13:33.101,0:13:37.745
Each, each message has a little bit of the[br]network connectivity and then the next one

0:13:37.745,0:13:42.302
comes after it. And so you could have many[br]messages going at the same time. And a

0:13:42.302,0:13:46.462
real long message won't fill up the[br]network, fill up the connection forever

0:13:46.462,0:13:50.783
and ever and ever. So it and it also[br]allows, if you to break the message up

0:13:50.783,0:13:55.530
into small parts, they can flow over[br]different paths. The other thing that they

0:13:55.530,0:13:59.989
figured out was this notion of instead of[br]oop, oop, oop come back here, come back

0:13:59.989,0:14:04.233
here. Instead of using computers as the[br]intermediate stop points, because in store

0:14:04.233,0:14:08.482
and forward you could have a lot of[br]messages so you tend to store them on

0:14:08.482,0:14:13.413
disks. Whereas routers, these packets were[br]smaller individually than the entire

0:14:13.413,0:14:18.515
message and so they didn't need to store[br]them nearly as long and they didn't need

0:14:18.515,0:14:22.953
as much storage. So these are, routers are[br]just a form of computer, right? But they

0:14:22.953,0:14:27.563
were specialized for moving just data from[br]one connection to the other without long

0:14:27.563,0:14:33.404
term, without storing that data for a long[br]time. So, I like to think of packets as

0:14:33.404,0:14:40.781
postcards, letters and think of the Packet[br]Switching Network as the postal system. So

0:14:40.781,0:14:47.226
let's say, for example, I had a friend,[br]and his name is Glen, and I want to send

0:14:47.226,0:14:52.645
him a message. I want to send him a[br]message that's hello there, have a nice

0:14:52.645,0:14:57.543
day. But I have a limitation. I have[br]limitation. All I have is postcards that

0:14:57.543,0:15:02.963
it can, that can store ten characters on[br]them, and I have to send my message to

0:15:02.963,0:15:08.530
Glenn using only 10-character postcards.[br]And so, before Glenn and I part ways, we

0:15:08.530,0:15:14.073
agree on the following protocol: that I[br]will take the first ten characters of the

0:15:14.073,0:15:19.586
me ssage and put them on one postcard, and[br]then I will put an address from Chuck to

0:15:19.586,0:15:25.075
Glenn, and I'll put a sequence number. So[br]that says that hey, hey Glenn, here comes

0:15:25.075,0:15:31.067
a message, this is part one. Then we take[br]the next ten characters. And I mark that

0:15:31.067,0:15:37.203
as part two, from Chuck to Glen. And then[br]here's the third part, it's marked as part

0:15:37.203,0:15:42.532
three, from Chuck to Glenn. And, so, what[br]can I do now? Well, I walk out to my post

0:15:42.532,0:15:48.047
office box, and I send'em, I just stack'em[br]in. I might stack them neatly in order.

0:15:48.047,0:15:53.545
Now, they go through the postal system.[br]Like, they get dropped, some get dropped

0:15:53.545,0:15:59.288
on the ground. A couple of them get lost.[br]Or they end up on the wrong truck. They go

0:15:59.288,0:16:04.513
through Kansas City by mistake. Blah,[br]blah, blah, blah, blah, blah, blah. Blah,

0:16:04.513,0:16:09.612
blah, blah, blah, blah, blah, blah, blah,[br]blah, blah, blah. But, you know, some days

0:16:09.612,0:16:14.653
later. They start arriving at Glen's[br]house. And so, Glenn goes out to his post

0:16:14.653,0:16:19.834
office box, and he gets a message. It's[br]hello ther-, and it's sequence number one.

0:16:19.835,0:16:25.074
So it looks like Chuck is going to send me[br]a message, and I've got the first part of

0:16:25.074,0:16:29.785
it. That's pretty cool. So then he goes[br]out the next day, and out comes, nice day.

0:16:29.785,0:16:34.602
But this is #three. So, because I've[br]numbered them, Glenn knows that there's

0:16:34.602,0:16:39.569
some missing bits, right? So Glenn just[br]can hold on to them, and leave a little

0:16:39.569,0:16:44.311
space on his kitchen table for what he[br]hopes to be message number two. And so

0:16:44.311,0:16:49.007
message two finally comes out. And now[br]Glen is capable of saying, "Looks like I

0:16:49.007,0:16:53.922
got the whole message and I can reassemble[br]them. And, surprise, surprise. I have just

0:16:53.922,0:16:58.363
sent him. With a lot of effort in three[br]packets. Hello there! Have a nice day."

0:16:58.363,0:17:03.561
And so this notion of breaking the message[br]into packets, labeling each packet with a

0:17:03.561,0:17:08.750
sequence number, and then sending them to[br]this network that can take multiple paths,

0:17:08.750,0:17:13.173
You can even have a situation where the[br]you know, the message would go across one

0:17:13.173,0:17:17.975
link, it would get lost and then it would,[br]you know go across a different link. So

0:17:17.975,0:17:21.961
you have ways of recovering. You can[br]recover the messages. We'll talk about

0:17:21.961,0:17:28.275
that later as well. So this ends up with a[br]sort of a structure that has these

0:17:28.275,0:17:36.238
computers that are specialized routers in[br]the middle. And the routers have multiple

0:17:36.238,0:17:43.091
connections. And if we take a campus, for[br]example, and the campus has some computers

0:17:43.091,0:17:47.484
and we have high-speed networking on this[br]campus. We have some, you know, stuff in

0:17:47.484,0:17:52.213
our offices on the campus, and then we[br]have some stuff in the machine room and we

0:17:52.213,0:17:56.461
talk to these things. And then, somehow,[br]our entire campus has a little spicket to

0:17:56.461,0:18:01.420
the outside world and this is our, sort[br]of, campus router and we get this router,

0:18:01.420,0:18:05.672
and then there are, sort of, intermediate[br]routers that are inside the network. And

0:18:05.672,0:18:09.688
if you sort of look at a router, a router[br]sort of simply forwards traffic and the

0:18:09.688,0:18:14.706
traffic now is these small packets, rather[br]than whole messages, so you don't need a

0:18:14.706,0:18:18.797
disk drive on these, on these routers.[br]There's no disk drive on these routers, so

0:18:18.797,0:18:22.689
that they just kind of grab a packet and[br]they forward it. And the systems are

0:18:22.689,0:18:27.084
trained. And the software does not[br]overflow the network. We'll talk about

0:18:27.084,0:18:31.968
that later, much later. And so these[br]routers have these real simple view of the

0:18:31.968,0:18:36.594
world, they've got some incoming traffic,[br]they've got some outgoing traffic,

0:18:36.594,0:18:41.537
outgoing traffic. And so they just grab[br]and forward. It's like a intermediate

0:18:41.537,0:18:47.240
postal spot, right? They, they grab big[br]thing of. Postcards and books. Send them

0:18:47.240,0:18:53.696
to the right place and, and they get where[br]they do and so eventually the data gets.

0:18:53.696,0:18:58.385
Getting a little sloppy, getting a little[br]messy, here. Eventually, the data sort of

0:18:58.385,0:19:03.304
is broken up, finds its way to the other[br]end, and then dumps out in some campus

0:19:03.304,0:19:07.743
local area network and then somebody sees[br]the data on the far end, okay? And so it

0:19:07.743,0:19:12.658
might different, take different routes,[br]you know? It might get lost that might

0:19:12.658,0:19:17.089
crash and then it has to get sent again on[br]a different route. And so these things,

0:19:17.089,0:19:21.097
these little pieces, these little[br]postcards, find their way through the

0:19:21.097,0:19:26.086
series of routers. And we can, we both can[br]see sort of like a, a whole campus being

0:19:26.086,0:19:32.006
connected. We can see individual folks who[br]are, buying, buying some dial up through

0:19:32.006,0:19:37.030
cable or DSL, and at some point we like to[br]represent this whole thing. Here is this

0:19:37.030,0:19:42.034
big cloud, this you don't worry about the[br]detail inside here. Call that the cloud.

0:19:42.034,0:19:45.757
We'll see it in the future slides it`s[br]just a cloud, a white, fluffy cloud. That

0:19:45.757,0:19:49.498
means that we are trying to hide the[br]detail. But in there it`s just a bunch of

0:19:49.498,0:19:53.265
things that are connected. In a way it`s[br]not that different in the store and

0:19:53.265,0:19:57.974
forward network, except for the fact that[br]every message is tiny, so it doesn't clog

0:19:57.974,0:20:01.665
the whole network up, which means that[br]routers don`t have to have a lot of

0:20:01.665,0:20:06.470
intermediate storage to hold on to these[br]packets in flight. And it also means that

0:20:06.470,0:20:10.747
every packet can take a different path and[br]if things get loaded up, they can

0:20:10.747,0:20:14.941
dynamically move. And so. Here's just sort[br]of an example problem to solve. If you

0:20:14.941,0:20:19.706
think about it, these routers have a very[br]limited view of the world. And there are

0:20:19.706,0:20:24.795
hundreds of thousands of routers around[br]this world right now. And they don't know

0:20:24.795,0:20:29.210
the entire network, they kind of know the[br]lines that come in to them and the lines

0:20:29.210,0:20:33.866
that go out, just like a post office in[br]Kansas city doesn't know every address,

0:20:33.866,0:20:38.583
every house in the world. It just knows[br]the trucks that are coming in and the

0:20:38.583,0:20:44.556
trucks going out. And so these packets[br]that have to and from addresses can get a

0:20:44.556,0:20:51.133
little confused at times. So we won't[br]solve this but if, if we had a situation

0:20:51.133,0:20:55.981
where This particular packet would come[br]into a router, and it would route it here,

0:20:55.981,0:20:59.426
and then this packet would see it and then[br]it would route it this way, this packet

0:20:59.426,0:21:03.228
would see it and route it this way, this[br]packet would see that this router would

0:21:03.228,0:21:06.864
see it again and say, oh, I gotta route it[br]that way. And so we end up in this

0:21:06.864,0:21:11.938
situation where we would create a loop.[br]Okay. So this is the kind of technical

0:21:11.938,0:21:17.157
things they had to solve to keep these[br]things from going round and round and

0:21:17.157,0:21:22.954
round and sort of melting the network.[br]We'll talk more about that in a bit. So.

0:21:22.954,0:21:28.364
This was DARPANET. It was doing research[br]on these kin ds of problems. The kinds of

0:21:28.364,0:21:32.988
problems of, you know what's the best way[br]to do this? How big should packets be?

0:21:32.988,0:21:38.392
What should, how long should we wait until[br]we send a packet again? You know, this

0:21:38.392,0:21:43.428
kind of thing. And so that was our[br]research network. And the, that could've

0:21:43.428,0:21:49.654
gone on forever, it might've been a purely[br]military project, but. At the University

0:21:49.654,0:21:57.237
of Illinois - Urbana Champagne - folks[br]started to think about super computers and

0:21:57.237,0:22:03.506
starting all the way back to Bletchley[br]Park, science was enhanced by the use of

0:22:03.506,0:22:08.802
computations. And so as the 70's and 80's[br]were happening, all these scientists were

0:22:08.802,0:22:13.037
sort of like, "Wow, I can do better[br]physics. I can do better chemistry. I can

0:22:13.037,0:22:17.607
do better material science. I can invent[br]new plastics. I can do pharmacy. I can do

0:22:17.607,0:22:22.741
all kinds of things. With computers. And[br]so what happened was is everyone started

0:22:22.741,0:22:27.448
asking the government. For money. For[br]computers. It's like, "I need a bigger

0:22:27.448,0:22:32.077
computer. And if I, if I had this bigger[br]computer. I could do research." Matter of

0:22:32.077,0:22:35.972
fact. I was part of all this. Matter of[br]fact I wrote a book. High Performance

0:22:35.972,0:22:40.569
Computing. Here's the book. That's kind of[br]what I did before I became an internet

0:22:40.569,0:22:45.606
guy. These are beautiful things. Here's,[br]this isn't, was my baby, I never got this.

0:22:45.606,0:22:50.965
This is like about $8,000,000, it's not[br]small like this, this is a model of a

0:22:50.965,0:22:56.575
Convex C3800 supercomputer. And each of[br]these was the size of a refrigerator, it's

0:22:56.575,0:23:02.109
slightly taller than me. I would be about[br]this tall, right here. And each of these,

0:23:02.109,0:23:08.160
I think this is like, like I said, like[br]$8,000,000 or something. And I wanted one

0:23:08.160,0:23:13.164
just for me. And so the problem is, is[br]that, you know, I'm a nice guy, and I'm

0:23:13.164,0:23:17.093
probably worth $8,000,000 of the[br]governments money without a doubt, but not

0:23:17.093,0:23:21.451
that the government didn't always think[br]about that. So we couldn't all have out

0:23:21.451,0:23:26.994
own personal computer, or at least our own[br]personal supercomputers. Today, of course.

0:23:26.994,0:23:32.807
This has about as much power as this, but[br]this is not a history of computers.

0:23:32.807,0:23:38.542
Computation. The iPhone is as powerful as[br]this thing, it literally with abou t as

0:23:38.542,0:23:42.625
much storage, But what happened was, is,[br]all these scientists would say give me,

0:23:42.625,0:23:45.860
give me this supercomputer. I need a[br]supercomputer to do this, I need a

0:23:45.860,0:23:50.216
supercomputer to do that. And the National[br]Science Foundation said oh, hmm, well, why

0:23:50.216,0:23:53.446
don't we just buy a few of these[br]supercomputers and put them in these

0:23:53.446,0:23:57.081
supercomputer centers and then let people[br]connect to them. And then make people, and

0:23:57.081,0:24:01.700
make it so they could share, so we didn't[br]have to give every single scientist one of

0:24:01.700,0:24:08.181
these things. And so. The notion that we[br]would create a network to connect these

0:24:08.181,0:24:16.069
things, again, seems completely logical[br]today, but in 1981, 1982, 1983, it wasn't

0:24:16.069,0:24:23.154
entirely the most logical idea. And of[br]course, the telephone companies might have

0:24:23.154,0:24:28.523
something to say about that and so the[br]next person that you're going to meet is

0:24:28.523,0:24:34.029
Larry Smarr from NCSA, the National Center[br]for Supercomputing Applications. And Larry

0:24:34.029,0:24:39.440
Smarr was one of the early innovators that[br]sort of realized that we had to build

0:24:39.440,0:24:44.684
computational infrastructure and internet[br]computational infrastructure. And did a

0:24:44.684,0:24:50.098
lot of work to convince the federal[br]government that this is something that we

0:24:50.098,0:24:54.020
should do. And so let's go ahead and meet[br]Larry Smarr.