What Happens When You Click a Link? - Computerphile

[@Neueregel]

Good info. Please, consider asking Dr. Mortier about VPN in the future. He seems very knowledgeable. I have a hard time understanding how VPN networks work.

[@underweightHater]

i love the graphics in these videos. they're so unusual and handcrafted looking (in a computery way)

[@xanokothe]

The background it's even more complex when we talk about highly distribuited systems. At same time we could (programmers, engineers, scients) make it incredibly easy for end user and, of course, highly lucrative for a lot of sectors.

[@MrGiovanniOSFP]

Just awesome...One of the finest channels. Thanks a lot for this content.

[@darkflamesquirrel]

Fascinating. I hope this is covered in a later video: "What happens when you click on a Zelda?"

[@VonZombus]

I always come back to this 🙂

[@1954BJohn]

Explained so clearly - Thanks!

[@pratiyushprakash6509]

This is so well explained. Great video.

[@aka-vm]

here because of Dr. Chuck

[@SumitKumar-fn3gj]

Love this. Beautifully explained

[@Manabender]

I just noticed this... In google chrome, the URL bar color-codes the URL. The schema is in green, the domain is in black, and the path is in grey.

[@TheAdriyaman]

I know. And I think the green is in red if it is an insecure connection.

[@maw570]

Dr Richard Mortier is a G

[@Myx0]

HTTP/1.1 solves the problem (among others) of having multiple websites sharing a single IP address... The domain name is passed to the server (along with the path), allowing the server to route the request to the correct website.

[@tabularasa0606]

Hurray for virtual servers!

[@ShynRou]

tabularasa0606 Pretty much all vServers already have an own ip if not multiple.

[@FabiClawZ]

ShynRou You understood him wrong he doesn't ment virtual root server on hardware . I think he meant virtual instances of the Webserver software

[@bingebammer9927]

WHAT YEAR IS IT?!

[@tabularasa0606]

It's 1492, why do you ask?

[@SinthTeck]

When you click a Link, he gets mad and starts throwing swords at you

[@OGTunny]

this is actually perfect i love this

[@yijiang7779]

Very informative and helpful video, thanks for sharing!

[@zenzylok]

The complexity behind one click is quite amusing.

[@AliAlHabsiAA]

Really liking these kind of videos. Keep it up.

[@DogsBAwesome]

It does not happen much these days but waiting for some frigging advert to be downloaded used to hang or slow down the loading of the page, really annoying.

[@limitless1692]

great I always was curious about what happens behind a link

[@TheKesmond]

Protocol stack. think this is what you mean as complex, lots of things happen.

[@Zhab80]

Hmmm, no mentions of ports anywhere. Kinda important... How does the computer knows that this webpage is meant for the 3rd tab of chrome rather than your 2nd tab of firefox ? Ports. How does the computer knows that this url is meant to connect to a webpage server rather than a mail server or a video game server ? Ports. By the way... ports is what TCP actually handle. The IP protocol is the one that handle IP address (I guess that is why they are called IP address), You can't get far with just one of these, which is why you usually hear about TCP/IP in pair like that.

[@karlosdaniel6537]

It's just a summary video, or do you want him to talk about Sockets, TLS, Proxy Servers, Cookies, etc... there is so many things happening that is impossible to cover all of it in a video of 10 minutes.

[@SevenDeMagnus]

How many combination (or it that permutation?) can those 3 parts makes? Would there be a time when all addresses are taken and there's no combination or permutation that can be made, anymore? Thanks.

[@Tokkemon]

Very nice and informative video!

[@outofthebox7]

When I press a link, does the final product (some page) download to my computer, or do I see it on the other person's computer? Sorry, I'm learning.

[@grn1]

You may have already found the answer by now but in case someone else comes along. All of the data needed to display the page is either downloaded to your computer or sometimes streamed in. A basic HTML page is going to be fully downloaded and interpreted on the local system, in fact different browsers may display the page slightly differently (most everything nowadays is based on the chromium engine though so it usually looks the same on every system, mobile devices sometimes get a different version of the page but that's up to the browser to send a request for a mobile version and the server to actually have one to send). Streamed content is usually downloaded in chunks though it could also be a continuous stream (not sure on the terms). The continuous stream would be like a (AM/FM) radio channel where if some data is lost it never gets resent while the method most players will use is to download small chunks. Either way a stream is not fully hosted on your system but it is being downloaded as it goes. Hope that made sense.

[@Suav58]

When I was telling kids about it I was always using full names: Transfer Control Protocol, Domain Name Server, Hypertext Transfer Protocol. It is easier, much easier, to get the acronym when you know the full name and the meaning, than the other way round. An advantage of a "living" class is, that one can keep pupils involved by asking questions. The easiest and surest thing is to make it graphical: write some pseudo-DNS as a table, have some bits of paper with proper structure as packets. I takes a whole lesson, rather than 9:50, but it is worth it. Kids tends to know it well afterwards. Pity, that due to my convictions I am not allowed to teach any more.

[@intenzi69]

your convictions?

[@Suav58]

@@intenzi69 I am a communist.

[@maeon3]

I don't like how he takes about 1.5 minutes just to rephrase the question in about 25 different and superfluous ways.

[@cosmicWil]

I have a question that maybe you guys can answer. Let me explain. If you take a picture of a computer screen with your phone, lines appear on the image( obviously). However, the interesting part is when you zoom in and out of the image the lines begin to move and grow larger and then smaller again. It almost seems like a wave moving across the image. I have tried it on a few different android phones and it always happens. Can anyone explain why its happening?

[@Kantenkugel]

Wouldn't the very first thing be to check the protocol and associate the link with the correct program for a link of that protocol?

[@sofia.eris.bauhaus]

yes, it would. it's just that it's by far the most common behind a link on a website..

[@ZT1ST]

+Kantenkugel Not if the browser or connection's lazy - not to mention what would happen if you don't have a program set for that link. Older browsers may not actually know what to do when you click a "tel:" link, for example. Also, depending on the path in question, you may not get something back that's relevant to the program that clicks that link - so when your computer gets back a .docx file, for example, it'll then ask if you want to download it or open up a relevant app with the file it got, or outright automatically download it and let you know where it is.

[@arthare]

So would be faster to connect to a webpage by directly using the IP instead of the domain name? Or would it just trigger a DNS process asking what 10.0.0.1 is?

[@jakuba8]

in theory yes, faster, another advantage of this is that you could connect to the desired webpage even if the DNS is down for whatever reason, for as long as everything else works

[@sofia.eris.bauhaus]

also you browser might prefetch domains on the pages you visit.

[@MidnightSt]

yes, about 40 miliseconds or so faster.

[@sunday87]

Or you might not even get what you want because of a thing called VHosts. There might be multiple domain names pointing to the same IP adress, e.g. if multiple websites reside on the same server. The domain name is repeated in the GET request so that the host may give you the correct website that you asked for. If you used the IP adress only, the host will, if at all, send you its configured default website and that propably won't be what you wanted to get.

[@possiblydavid]

sunday87 Right, that's exactly what I was going to say. My webserver (server that I host from home) uses virtual hosts, so if you want to access one of my sites without using DNS, (something I do before buying a domain name while I set things up) you would need to add the dns entry to your computer's hosts file.

[@kbrandhawa2195]

why the internt speed differ like some connection are fast n some are slow to appear the web page??

[@SquareKiteGaming]

What about if it's slower or faster when a link is is loading to refresh it in a second tab?

[@PiotrBaczek87]

Hey ***** , maybe you could make a video about Google's material design and it's impact on creating websites? CSS's are cascading stylesheets but with material you can enclose style just to current element with shadow DOM. How is it done?

[@cybermonterey5087]

The whole thing could be explained in 3 minutes. No need to be wordy for the three way handshake... But a nice British accent:)

[@stryp]

The "hosts" file doesn't have an extension (no .txt). It just caught my attention for a bit. :)

[@richardmortier9614]

not on modern unix systems. but, to quote (eg) wikipedia, "Originally, a file named HOSTS.TXT was manually maintained and made available via file sharing by Stanford Research Institute for the ARPANET membership, containing the hostnames and address of hosts as contributed for inclusion by member organizations."

[@user-or7ji5hv8y]

What happens if you click on a link in a spam? Can something get installed on your laptop automatically, without your authorization?

[@Benimation]

What happens when 2 people register the same domain name at about he same time?

[@goommenter]

I got lost around 5:30 and got back in the end. So what did happened?

[@PaulaDima]

Fascinating

[@jamesmclennan6436]

Does that mean adblock software just denies any three-way handshakes with domains outside the one you want?

[@trevinbeattie4888]

Something like that; it depends on how the software is implemented. The one I use (AdBlock Plus) maintains a blacklist, where any host on that list is blocked; all others are allowed. Many sites these days have a such surprising amount of content loaded from 3rd-party servers that they wouldn't work at all if the browser blocked all domains but the parent. (I also use NoScript to block all code from 3rd-party domains, for safety.)

[@tabularasa0606]

An URI also has a port, the URI shown only works for the default port.

[@fluffyflextail]

"It's a complicated system" does imply that it's one system, that is were you're wrong :O ..... Rather make it somewhat more plural as in systems

[@tomazaswx]

To click, or not to click, that is the question -

[@raphaelhuber1]

I wonder: If you would construct the internet again, TODAY. Would this all be done in an easier way?

[@Qbe_Root]

I don’t think it would. You can’t choose your IP address, so using it directly as a domain name would be very limited (though technically, an IPv6 address could be represented as 16 characters or 21 alphanumeric characters). The syn-ack is required for the server and the client to know where they should send their data. To avoid fetching multiple files at once, it’s possible to write a whole web page in one single file, but I don’t think you could add pictures (let alone sound or videos) and the page would be a pain to develop. Most of the linked files are already cached by your browser or on the same server as the page itself anyway…

[@cat6024]

I clicked a link on youtube am i in danger.

[@0SL2]

What system uses Host.txt????

[@kd1s]

Oh I know how it works - and how to diagnose issues relating to it. When you've been in the I.T.field for as long as I have you tend to know this by rote.

[@Zandonus]

I vaguely remember from a lecture that the syn-ack's are where all sorts of nasty things can potentially happen. Pretending to be someone you're not, the haxxor pretending to be the server, packet loss.

[@Qbe_Root]

Well, it’s pretty much the step where the server learns who is the client and vice-versa, so it’s the best occasion for a hacker to get in…

[@SkukS]

Qbe Root How would the server learn who the client is with the syn-ack?

[@Qbe_Root]

Skux Deluxe My bad, the client and the server actually tell each other the sequence numbers they’re going to use. But that’s the only occasion for a hacker to find out the right sequence number to impersonate the client (or the server)…

[@Salad360]

Dr. Mortier how do we protect ourselves from Lizard Squad?

[@Zantorc]

Except 10.0.0.1 is a private IP so it's a non-routable address - it will never get onto the internet. The 10.X.X.X range is for use on local area networks.

[@BohonChina]

can you make a video about how does google CDN or some other CDN work?

[@karlosdaniel6537]

Yes, I know I'm 7 years late but in any case the last edition of Computer Networking: A top-down approach by Jim Kurose and Keith Ross discusses about CDN, specially for stored-video streaming services. In short, the Google's private network uses DNS redirect to intercept and redirect the users requests to the server of smaller RTT between the user and the server. It also places servers in both IXPs and access ISPs, uses a pull-caching strategy for replicating the KZbin videos in the CDN servers and the processing work of the videos is done in the few Google's data centers around the world.

[@hushedupmakiki]

***** If there is a syn-syn ack-ack is the limiting step, how does a DDOS attack work?

[@Isaac_Roberts]

a little late but this may sorta answer your question. So your pc sends a SYN to the target system, then the target system replies with a SYN/ACK. Well now you send a ACK right? no. Now you send another SYN and just keep on repeating that thousands of times. That is a dos (denial of service) attack because that was only one host. Now lets scale it even larger now you have thousands of hosts, called a bot net, all controlled by one computer that tells them to attack a specific target now becoming a DDOS (distributed denial of service) attack. Thats the basics theres a lot more to it lol.

[@sofia.eris.bauhaus]

SPDY and QUIC ftw! :D

[@sSunbeamM]

when a link does not work, i click as fast as i can, to annoy the internet until it's willing to do what i want!

[@niveks_]

How come the DNS servers never seem to get DDoS, even unintentionally by users?

[@MidnightSt]

they do, but: 1. there's so many of them in such a hierarchy you usually don't notice. first, your ISP's DNS is used, if it doesn't know or doesn't respond you can get routed to the one-higher DNS, etc. Also, I bet there's some caching so it's not required that the reply come from the DNS, it's possible that almost any server can look at the dns request and reply instead of forwarding it to the destination (but i'm not sure about this) 2. as you have noticed from the video, it's a one-time thing, you send a request, you get or not get an answer. there's not much info for you to determine if DNS got DDoSed, or if it was the site's server itself, as opposed to "i'm already on a site, i attempt an action (submit comment, open an article), the site fails to load" , where it's a clear failure you can see and assign to the site (usually correctly) even without any knowledge of how the web works.

[@SkyrimHod]

There have been a few attempts to DDoS the root DNS servers, but it would be extremely difficult to do so effectively. First of all, keep in mind that these servers are DESIGNED to be receiving huge number of requests constantly from all over the internet. They will have the capacity to handle several times more than what they actually expect to be receiving most of the time in case there is some sudden huge surge in demand. It would take an incredibly huge attack to bring them down. Then there's the fact that there are nominally 13 separate servers that would need to be hit simultaneously. In reality, it's actually over 100 physical locations scattered around the world. And you can bet that they have some of the top experts on cybersecurity working for those sites to deal with any attacks. I'm sure they have firewalls set up to stop anything that's not a legitimate dns request long before it even reaches their servers. And finally, most dns requests never even need to go to those root servers anyway since ISPs will be running their own more local dns servers and caching the information. Even if all the root servers were taken down for a short time, this would be probably be enough to handle most requests for a while.

[@RealCadde]

In line with what SkyrimHod says, one major reason the DNS network doesn't get DDOS'ed is because the ISP's cache the requests for as long as they have been set by the responder (the other DNS server), it's stored in the TTL (Time To Live) parameter of the response. It is of course entirely up to the local DNS to adhere to this TTL value but generally ISP's and all other (at least trusted) entities of the internet do adhere to the terms set up by others. So, if you repeatedly ask to resolve (which implies you have also disabled your machines local cache of DNS records) an address, all you accomplish in doing is overloading your nearest DNS and even then there's most likely a filter set up where you are only allowed so many requests before the server ignores your traffic, causing a minimal effect on it and your ISP investigating or outright stopping your service until you can resolve your abuse issues... Even when the issues weren't intentional from you, say you got a trojan that makes you part of a botnet. In simple terms, any traffic that looks suspect (not that your ISP actively monitors your traffic, but passively through firewall rules etc) may/will trigger a local block first. Especially on major ISP's that have some degree of security in place. Both for you and for the internet as a whole. Beyond that cache and the offender being ignored in one way or another, the DNS network is multilayered. Your ISP caches the results, the provider for your ISP (yes, they most likely have a provider themselves) caches the results and each step along the way caches the results. Which is why, when you update a routing for a specific address they ask you to wait at least 10 minutes before expecting the change to happen at your place. There's just so many layers of DNS'es to propagate through before a change actually gets cached. Which happens more frequently than the default refresh interval. Simply put, DDOS'ing the DNS network is about as difficult as singlehandedly overloading the entire internet with your bogus traffic. Which has happened before mind you, just not to the DNS network.

[@RealCadde]

An addition to my previous response, you can of course specify a DNS server to be used rather than the default one of your ISP but attacking one DNS server is not going to have much effect. The more important a function is, the more redundancy it has. Attacking one major DNS server (or two... three or even four etc) will just overload those servers. There are plenty more with enough capacity to answer in a reasonable amount of time and as i pointed out, DNS entries are cached on all DNS servers globally. Once cached, the MAIN servers are not affected by the "non attackers" and the "non attackers" completely bypass the affected DNS server(s).

[@Hemulen40]

It may sound daft...but I have no idea what yer talkin about ?

[@ping1509]

Why is there no subtitle...I can't hear you clearly

[@ping1509]

为什么没字幕？？英文字幕也可以啊。。。

[@rattlehead999]

It's simple, yet so long and hard to remember :/

[@Pyotyrpyotyrpyotyr]

No brown paper?

[@trevinbeattie4888]

Of course if the Web site is using any kind of JavaScript on its links, this whole description goes out the window. (Grumble)

[@thuokagiri5550]

looks like Dr Mike Pound from 2045

[@SkukS]

Hosts.txt??? No .txt extension on that one buddy.

[@richardmortier9614]

not on modern unix systems. but, to quote (eg) wikipedia, "Originally, a file named HOSTS.TXT was manually maintained and made available via file sharing by Stanford Research Institute for the ARPANET membership, containing the hostnames and address of hosts as contributed for inclusion by member organizations."

[@JulianDrangosch]

Who is here from Django For Everyone?

[@nekkun831]

This is like the first thing you learn in networking class lol

[@SquareKiteGaming]

Who the hell maintains DNS anyways?

[@Hahahawhatsup]

Anyone else think he looks like a nerdy Simon Pegg?

[@Ryan-9000]

He kinda looks like Frankie Boyle

[@laetitiabaschenis-lefebvre5513]

Hello I'm French but I do understand English, well most of the time. Expect that in this video, I just couldn't understand everything and I wonder if it's not because of the weird accent or perhaps he just doesn't speak clearly. I don't know. Can someone tell me if I'm just out of my mind, and in this case, I'll see someone, or confirm that he has a strange enunciation ?

[@TheMrTape]

It's his accent, I think it's specifically a Nottingham accent.

[@RMoribayashi]

He was speaking fast, softly and with a _very_ low pitched voice. Some post-processing to emphasize his consonants would have helped. My hi-fi speakers handled it but I had to turn up the volume. Telco line also are notorious for mangling bass voices like his (and mine). I often have to raise the pitch of my voice to be understood on the phone.

[@MrWadeBarrett]

Your English is just bad...

[@maagiapall]

TheMrTape I had absolutely no problems understanding the accent.

[@MidnightSt]

Hello, I'm native Slovak and I had no trouble understanding. I actually like this kind of accent/speaking style, it seems very pleasant and easy to understand at the same time. But I'm an IT guy, so for you the problem might also be with the IT terms he uses, they tend to throw people off and make them unsure about whether they understand, make it hard to focus on the words after those terms as your brain doubts whether it was really english and falls back to "let's figure out what language it is first, then we'll try to parse the meaning"...

[@muhammadhassaan3785]

well explained watching too late now the web 3.0

[@fleetwoodsucks]

omg dont you ever buy new clothes?

[@klaxoncow]

This is not a live broadcast. They almost certainly filmed all these videos together on exactly the same day. Hence he's got the same shirt on from the previous video because, in reality, this was filmed immediately after the last one was. Just because days or weeks pass for you between video uploads does not mean that what you're seeing in those videos has transpired on the same time scale. (And even if weeks did pass between videos, then another simple explanation is that he has worn plenty of other shirts in the meantime but, by sheer coincidence, he just so happened to be wearing the same shirt on these different days of filming.)

[@fleetwoodsucks]

KlaxonCow The shirt is from 2006.....

[@zelloxy]

Normally Computerphile is awesome, but this was quite basic

[@eggfun1467]

(_yawn_) This video is so boring