Chapter 1: Computer Networks and the Internet

Welcome to Last Minute Lecture.

This free chapter overview is designed to help students review and understand key concepts.

These summaries supplement not replaced the original textbook and may not be redistributed or resold.

For complete coverage, always consult the official text.

Welcome to The Deep Dive, the show that cuts through the noise to bring you surprising, digestible insights.

Today, we're plunging into a system so vast it's hard to even, well, imagine the internet.

It really is staggering, arguably the largest engineered creation in human history.

Totally, connecting billions of us.

And now, like all these things, smartwatches, thermostats, even cars, it's everywhere.

Our mission on this deep dive is to give you a genuine shortcut to understanding this huge, complex system.

We're digging into its core principles using a foundational textbook as our guide.

The goal really is so you can grasp not just today's networks, but also get a feel for what's coming next.

Exactly.

We're taking a top -down approach today, starting with the apps you use every day, and then peeling back the layers.

Getting down to the nuts and bolts.

Right.

How data actually moves, what makes it slow down sometimes, and even how it can be attacked.

So let's unpack this.

What is this massive interconnected thing we call the internet?

That's the perfect place to start because you can look at it in a couple of ways, really.

From a nuts and bolts view.

The hardware view.

Yeah, the hardware.

It's basically a global network of connected computing devices.

Used to be mostly desktops, servers.

But now it's everything.

Phones, tablets.

Billions of phones, yeah.

Plus gaming consoles, TVs, security systems, cars,

all sorts.

We call these hosts or end systems.

They sit right at the edge of the internet.

Okay, so our devices are the edge.

How do they actually plug into this global thing?

They connect through communication links.

Think of them like roads for data.

Different kinds of roads.

Definitely.

You've got your old school copper wires, coaxial cable like for your TV, super fast optical fiber, and of course, radio waves for wireless.

Wi -Fi, cellular.

Exactly.

And each has a different speed limit, basically.

A transmission rate measured in bits per second.

And when your device sends data, it chops it up first.

Into smaller pieces.

Into manageable chunks called packets.

Each packet gets a little header, like an address label.

Ah, and these packets then go through packet switches, right?

I remember that analogy about trucks on highways.

It's a great analogy.

Packets are like the trucks carrying digital stuff, the links to the highways.

And the packet switches, they're the intersections.

Threatening traffic.

Precisely.

A switch takes a packet coming in, looks at its address label, and sends it out on the right road towards its destination.

In the core of the internet, we call these routers.

Closer to you, maybe in your office network, they're often called link layer switches.

So all these separate networks, my home network, the office, my phone connecting, how do they join the main highway system?

That's the job of internet service providers, ISPS.

Your home broadband company, your university's network, the mobile carrier for your phone.

The on -ramps.

Got it.

And the internet itself isn't just one big company, it's a network of networks.

Your local ISP connects to bigger regional ISPs, and they connect to even bigger national or international ones, these tier one providers.

They all interconnect.

Wow.

That sounds like it needs a lot of rules to work without just clashing into chaos.

Oh, absolutely.

And that's where protocols come in.

Protocols are just agreed upon rules.

They control how information gets sent and received.

Like a language everyone agrees to speak.

Kind of, yeah.

The most important ones are TCP, transmission control protocol, and IP, internet protocol.

You always hear them together.

TCPIP.

That's the core language of the internet.

And who decides these rules?

Mostly an organization called the Internet Engineering Task Force, the IETF.

They develop these open standards, publishing them as RFC's requests for comments.

There are thousands.

It's like a massive public instruction manual that ensures, your laptop in London can talk to a server in Sydney without any issues.

That's the plumbing sorted.

But the internet isn't just pipes, is it?

It's where we do stuff.

What about the services view?

Right.

From that angle, the internet is an infrastructure that provides services to applications.

All the things you actually use, email, the web, streaming video, social media, games.

Those run on our devices, right?

Not in the network itself.

Exactly.

That's key.

The applications run on the end systems, your phone or computer.

The packet switches in the middle, the routers, they just move the data packets.

They don't really care what's in them.

So if the apps are on my device, and maybe on a server somewhere else, how do they actually talk to each other across this huge network?

They use something called a socket interface.

Think of it as a doorway or a mailbox for applications.

It's a standard set of rules a program uses to send data specifically to another program on a different machine.

Like sending a letter.

Perfect analogy.

If Alice wants to send a letter to Bob via the post office, she can't just chuck it out the window.

She follows the rules.

Envelope, address, stamp, mailbox.

The socket interface is the internet set of rules for programs sending data to each other.

Okay.

Okay.

It's still kind of amazing how these digital conversations happen across continents.

What are these rules, these protocols at their very core?

How do they work?

Well, think about human protocols first.

We use them all the time without thinking.

If you want the time, you might say hi.

The other person says hi.

Then you ask.

There are specific messages, actions, and expected sequence.

Like a little script we follow.

Exactly.

A script or a dance.

If one person doesn't follow the script, communication breaks down.

In networking, the people are hardware or software things, computers, routers, phones.

Every interaction online is governed by a protocol.

Give me an example.

Okay, getting a webpage.

Your browser first sends a connection request message to the web server.

The server says, okay, connected.

Then your browser sends a message saying, send me this specific page.

The server then sends the page data back.

Ah, so it's a defined sequence of messages and actions.

Precisely.

A protocol defines the format of the messages, the order they're sent in, and the actions taken when a message is sent or received.

It's the choreography that makes the internet work.

Understanding networking is really about understanding these different choreographies.

Fascinating.

Okay, let's move from the big rules back to the edge.

We talked about end systems, our devices.

What else is at the network edge?

Right.

End systems or hosts are where the applications live.

We usually differentiate between clients that's typically us with our laptops, phones, and servers.

The powerful machines doing the heavy lifting.

Yeah, storing websites, streaming video, handling email, and most servers nowadays live in these enormous data centers.

I've heard about those.

Google, Amazon.

They have huge ones, don't they?

Unbelievably huge.

We're talking facilities with tens, sometimes hundreds of thousands of computers packed together.

They're the hidden engines behind pretty much everything we do online.

What do they actually do?

Well, three main things, generally.

One, they serve content showing you webpages, streaming movies.

Two, they handle massive calculations like for scientific research or data analysis.

And three, they provide cloud computing.

Ah, the cloud.

Right.

Think about a company like Airbnb.

They don't run their own servers all over the world.

They run their entire business on Amazon's cloud, on AWS.

That makes total sense.

Okay, so we have end systems, data centers.

How do we, at home, we're in the office, connect to this whole internet?

That's the access network you mentioned.

Exactly.

The access network is your on -ramp.

For home users, the common options are things like DSL and cable.

DSL uses the phone line, right?

Yep.

Existing telephone lines.

Speeds can vary, often faster downloads than uploads, and it depends how far you are from the phone company's local office.

And cable?

Cable uses the TV cable infrastructure.

It's usually a mix fiber optics to your neighborhood then coaxial cable into your house.

The key thing with cable, though, is it's often a shared connection in your neighborhood.

Meaning?

Meaning if lots of your neighbors are online, streaming movies, downloading big files at the same time.

My speed could drop, like rush hour on the highway.

Exactly like rush hour.

Everyone's sharing the same road capacity.

For potentially higher, more consistent speeds, there's fiber to the home, FTTH.

That brings fiber optic cable right into the house.

Directly.

Offering speeds in the gigabits per second range.

Verizon FIOS is a well -known example.

And there's a newer option emerging.

5G fixed wireless.

High speeds without needing a physical cable run to your house.

Interesting.

And inside buildings like offices or even homes?

For wired connections, Ethernet is still king, usually using twisted pair copper wires.

Speeds range from, you know, 100 megabits per second up to 10 gigabits or even more now.

And wireless.

That's Wi -Fi, of course.

Everywhere now.

Shared speeds, maybe over 100 megaps typically, but the range is shorter, maybe tens of meters.

And then for when you're out and about, there's wide area wireless.

3G, 4G LTE, and now 5G cellular.

If we zoom out again, all these connections rely on some kind of physical medium, right?

The actual stuff the signals travel through.

Absolutely.

That's the foundation.

We talk about guided media, where the signal follows a physical path.

Like twisted pair copper wire, cheap, common, used for phones and Ethernet.

It's twisted to cut down on interference.

Then there's coaxial cable, think TV cable, which can handle higher speeds.

And then the star, fiber optics.

Glass strands carrying light.

Tiny flexible glass strands.

Light pulses carry the data.

The advantages are huge.

Incredible speeds, immunity to electrical noise, very little signal loss over long distances.

That's why it's used for all the long haul stuff.

Under the ocean, the internet backbone.

Makes sense.

And unguided media, that must be wireless.

That's radio waves, yeah.

Propagating through the air or space.

Terrestrial radio covers everything from Bluetooth and Wi -Fi locally, to cellular networks covering wider areas.

And then you have satellite radio links.

Like for satellite TV or internet?

Exactly.

You've got geostationary satellites way up high, they stay over the same spot, but there's a noticeable delay because they're so far away.

About 280 milliseconds just for the signal to go up and back.

And then there are low earth orbit satellites, much closer, offering lower delays.

Okay, that covers the edge and the physical links.

Now let's dive into the network core.

This mesh of routers and links connecting everything.

How does data actually navigate this?

The core relies almost entirely on packet switching, as we mentioned.

Data gets broken into packets, and each packet finds its own way.

A really crucial concept here is store and forward transmission.

Store and forward.

Yeah.

It means a router has to receive the entire packet, every single bit, before it can start sending that packet out on the next link.

It can't start sending the beginning while the end is still arriving?

Nope.

It has to store the whole thing first, check it, figure out where it goes, then forward it.

So that adds delay, right?

Depending on the packet size and the link speed.

Absolutely.

That's called transmission delay, the time it takes to push all the bits of the packet onto the wire.

Longer packet, slower link means more transmission delay.

What happens if lots of packets arrive at a router wanting to use the same outgoing link?

Like a traffic jam.

Exactly like a traffic jam.

That's where queuing delays and packet loss come in.

Routers have buffers, like waiting rooms or queues, for each outgoing link.

So packets wait in line?

They wait in line if the link is busy.

The time spent waiting is the queuing delay.

And this delay can vary a lot, depending on how congested that link is.

And if the waiting room gets full?

If the buffer overflows because packets are arriving faster than the link can send them out, the router has no choice but to drop newly arriving packets.

That's packet loss.

Ouch.

So you want to avoid that?

Definitely.

Network engineers try to design things, so the average arrival rate is less than the link's capacity.

But bursts of traffic can still cause temporary congestion and queuing, or even loss.

Okay, so router store, forward, and sometimes queue or drop packets.

How do they know where to forward them?

Is there a map?

Kind of.

Every device connected to the internet has a unique IP address.

It's like a postal address for the digital world.

And the router reads this address.

Exactly.

When a packet arrives, the router looks at the destination IP address in the packet's header.

It then consults its forwarding table.

This table essentially says, for packets going to this range of addresses, send them out this specific link.

Like asking for directions at each intersection.

That's a great way to put it.

The router is like this helpful attendant giving you the next turn.

But these tables aren't programmed by hand usually.

How do they get built then?

Through special routing protocols.

These are protocols that run between routers, constantly exchanging information about the network's layout and traffic conditions, so they can automatically figure out the best paths and update their forwarding tables accordingly.

So that's packet switching.

But you mentioned another way networks can work.

Circuit switching.

How's that different?

Circuit switching is the older model.

Think traditional telephone networks.

The key difference is resource reservation.

Reservation.

Before any data flows, a dedicated end -to -end connection, or circuit, is set up.

Resources along that path, like a specific amount of transmission capacity on each link and buffer space in the switches, are reserved exclusively for that connection for its entire duration.

So it's like having your own private lane on the highway, guaranteed for your whole trip.

Exactly.

Whether you're actually using it every second or not, it's yours.

Packet switching is more like everyone sharing the public lanes, taking turns as needed.

How did circuit switching actually reserve that capacity?

Two main ways.

Frequency division multiplexing, FDM, or time division multiplexing, TDM.

FDM divides the link's bandwidth into different frequency channels, like radio stations.

TDM divides time into repeating frames, and each connection gets a guaranteed time slot within each frame to transmit.

But you said packet switching is dominant now.

Why?

Efficiency, mainly.

Computer traffic is usually bursty, lots of data, then silence, then more data.

Circuit switching is wasteful during those silent periods because the reserved circuit sits idle.

Packet switching allows many users to share the same link much more effectively, only using capacity when they actually have packets to send.

It's called statistical multiplexing.

So the trend is definitely towards packet switching.

Overwhelmingly.

Even voice calls over the internet now use packet switching.

Okay, so zooming back out, how does this network of networks actually fit together, all these ISPs connecting?

It's a really complex structure, honestly, driven a lot by business deals and geography.

But basically, you have this hierarchy.

Access ISPs connect to regional ISPs who connect to the big tier one providers.

Think of it like local roads connecting to state highways connecting to the national interstate system.

And money flows upwards?

Generally, yes.

An access ISP pays the regional ISP it connects to, and the regional ISP pays the tier one provider.

The tier one providers often have peering agreements among themselves.

Peering, what's that?

And you mentioned other terms like POPs and multi -homing.

Right.

A point of presence, or POP, is just a location, like a data center, where an ISP has routers that allow other networks to connect to its network.

Multi -homing is when an ISP connects to two or more different provider ISPs.

That's for reliability.

If one connection goes down, they still have another path.

Smart.

And peering.

Peering is usually when two ISPs at the same level in the hierarchy, often in the same geographic area, agree to connect directly to each other.

They exchange traffic between their respective customers without paying a higher tier provider.

Often, this is settlement -free, meaning no money changes hands if the traffic flow is roughly equal.

Where does this peering happen?

Often at Internet Exchange Points, or IXPs, these are physical locations,

basically big data centers where lots of different ISPs bring their routers and cables to interconnect directly.

It's like a big marketplace for swapping traffic.

And what about the huge content providers like Google, Netflix, Facebook?

How do they fit in?

Ah, they've changed the game significantly.

Companies like Google have built their own massive private global networks connecting their data centers worldwide.

So they don't just rely on the public Internet.

They try to bypass large parts of it.

They strategically place their servers close to users and often peer directly with access ISPs or meet them at IXPs.

This cuts down on the costs they'd pay to tier one providers and gives them much more control over performance, making your search results or videos load faster.

Fascinating.

Okay, let's talk performance.

We experience it as fast or slow, but what are the technical measures?

Delay, loss, throughput.

Crucial concepts, yeah.

When a packet travels, it encounters different types of delay at each router or node it passes through.

There are four main ones.

Four, okay.

First, processing delay.

Tiny amount of time the router takes to examine the header, check for errors, decide where to send it, microseconds usually.

Good tiny.

Second, queuing delay.

We talked about this, the time spent waiting in the buffer if the outgoing link is busy.

This one is highly variable, from near zero to milliseconds or even longer if it's congested.

Or the waiting line delay.

Third, transmission delay.

Also mentioned the time needed to push all the packet's bits onto the link.

Depends on packet size, L, and link speed.

Formula is LR, microseconds to milliseconds.

Got it, size over speed.

And fourth, propagation delay.

This is the time it takes for a bit to physically travel across the link's distance, D, at the propagation speed, S, which is close to the speed of light.

Formula is D, milliseconds usually, but can be significant for long distances, like satellites.

So transmission delay is about getting the whole packet onto the road.

Propagation is about travel time on the road.

Excellent way to put it.

Total delay at one router is the sum of all four.

Processing plus queuing plus transmission plus propagation.

And the total end -to -end delay is the sum across all the routers in the path.

Exactly.

Tools like TraceRoute help measure this.

It sends packets that make each router along the path, send back a reply measuring the round -trip time.

You can actually see the path your data takes and where the big delays occur, like that jump across the Atlantic on fiber.

Cool.

What about throughput then?

Is that just speed?

Throughput is the actual rate, in bits per second, at which you are receiving data at the destination.

Think of it as the effective speed you're getting for a specific download.

What limits it?

It's limited by the bottleneck link, the slowest link along the entire path between the source and destination.

If you have a super -fast connection at home, but the server you're connecting to has a slow link, your throughput will be limited by that server's link.

Or if lots of people are sharing a link somewhere in the middle.

Exactly.

If 10 people are downloading simultaneously over a shared 5 -millipedes link, each might only get around 500 kilobytes of throughput.

Which brings up an important point.

People often complain their internet is slow.

Why?

Well, very often the actual bottleneck isn't the huge internet backbone, which is incredibly fast.

It's frequently your access network, that connection between your home and your ISP.

That's the slowest part of the path.

Okay, that covers performance.

Now, unfortunately, you have to touch on the dark side networks under attack.

The internet's critical, but also vulnerable.

Sadly, yes.

Because it's so important, it's a constant target.

One big threat is malware, malicious software.

Viruses, worms, Trojan horses.

What can they do?

All sorts of nasty stuff.

Delete your files,

install spyware, distil passwords, bank details, social security numbers.

Or worse, they can turn your computer into a zombie.

A zombie?

Yeah, part of a botnet.

Thousands, even millions of compromised computers, all controlled remotely by an attacker.

They use botnets to send spam email or launch massive attacks, and malware can spread incredibly fast.

Scary.

What about those attacks you mentioned, denial of service?

Boss.

DOS attacks aim to make a service unavailable.

Three main ways.

One, send cleverly crafted messages to exploit a bug and crash the server.

Vulnerability attack.

Two, just flood the target's connection with so much garbage traffic that legitimate users can't get through.

Bandwidth flooding.

Three, overwhelm a server by making tons of fake connection requests.

Connection flooding.

And D -DOS is even worse.

Distributed DOS, yes.

Instead of attacking from one place, the attacker uses a botnet, thousands of zombies, to bombard the target from all over the world simultaneously.

Much harder to defend against.

Yikes.

What about privacy?

Can people eavesdrop?

Definitely.

That's packet sniffing.

On some types of networks, like Wi -Fi or older shared ethernet, someone nearby can potentially capture copies of all the data packets flying through the air or wire.

Including passwords.

If they're sent unencrypted, yes.

Passwords.

Personal info.

It's called sniffing because they're just passively listening.

Hard to detect.

And can attackers pretend to be someone else?

Yes.

That's IP spoofing or masquerading.

An attacker can create packets and put a fake source IP address on them.

They could make it look like the packet came from a trusted server or user.

So you might trust a malicious command.

Potentially, yes.

This is why authentication, proving who you really are, is so critical online.

You can't just trust the source address on a packet.

And this really hits on a fundamental issue.

The internet's pioneers, they designed it, assuming, sort of, that users would trust each other.

Openness was key.

But today, that trust model is broken.

We constantly interact with unknown people.

Unknown systems.

Securing this environment, built on an old assumption of trust, is maybe the biggest challenge in networking today.

It really is.

You know, a bit of history always helps put things in perspective.

Where did this all even start?

It goes back to the early 60s, believe it or not.

Back then, it was all about the telephone network using circuit switching.

But computers send data differently in bursts.

Researchers like Kleinrock at MIT, Baron at Rand, Babies in the UK, they all independently came up with the idea of packet switching around the same time.

And that led to ARPANET.

The very first large -scale packet switch network.

Yeah.

Funded by the U .S.

Department of Defense's ARPA.

The first node went live at UCLA in 1969.

Kleinrock tells a funny story about sending the first message.

They tried to type L -O -G -I -N to log into a remote computer.

They typed L, it worked.

Type O, worked.

Type G, and the whole system crashed.

So the first message ever sent was just LO, as in lo and behold.

Huh, brilliant.

So ARPANET grew, but other networks started popping up, too.

Lots of them in the 70s.

ALOHANET in Hawaii, using radio, satellite networks, commercial networks like TELANET, the challenge became connecting these different networks together.

Internet working.

Exactly.

That term was coined by Vint Cerf and Bob Kahn, who were developing the core protocols needed for this.

TCP and IP, they were initially combined, then split.

Bob Metcalf developed Ethernet around then, too, for local networks.

Then the 80s saw a huge boom.

A massive proliferation.

Host count went from hundreds to hundreds of thousands.

University networks like NSFNET became crucial backbones.

And the big milestone, January 1, 1983, Flag Day.

That's when ARPANET officially switched over to using TCPIP as its standard.

DNS, the Domain Name System, also came along, letting us use names like Google .com instead of raw IP numbers.

In the 90s, that feels like when it exploded into public view.

Totally.

The 90s were the internet explosion.

ARPANET was retired, NSFNET was commercialized, and commercial ISPs took over the backbone.

But the real trigger?

The World Wide Web.

Tim Berners -Lee.

Invented by Tim Berners -Lee around 8991, HTML, HTTP, the first browser and server.

Then Mark Andressen's Mosaic browser, later Netscape, made it graphical and easy to use.

Suddenly, there were killer apps.

Email was already big, but now web browsing, e -commerce, instant messaging, even early file sharing like Napster.

That's when it really hit the mainstream.

Followed by the dot -com boom and bust, of course.

Right, but the foundation was laid.

So what defined the internet in the 2000s and beyond?

Several huge trends.

First, broadband everywhere.

Cable, DSL, fiber made high -speed access common, enabling video streaming, YouTube, Netflix changed everything.

Second, wireless dominance.

Wi -Fi became ubiquitous and smartphones took off, driving mobile internet use way past desktop use.

Third, social networks.

Facebook, Twitter, Instagram created these massive networks on top of the internet.

And the big tech companies consolidating power.

Yeah, the rise of content provider networks, like Google building its own global infrastructure to deliver services faster.

And finally, cloud computing.

Companies moving their IT infrastructure onto platforms like AWS or Azure.

All these trends built on that packet -switched TCPIP foundation laid decades earlier.

So there you have it.

A whirlwind tour, really, of the internet's core ideas from the devices we hold to the global infrastructure and how it all came to be.

It really is a marvel.

We've touched on how packets move, what causes delays, the layered design, the constant security fight, and its incredible history.

And maybe that key takeaway to ponder is the shift away from that original design assumption of mutual trust.

Think about it.

How do we navigate?

How do we secure a global network where trust is no longer the default starting point?

It impacts everything we do online.

It absolutely does.

A huge question for all of us.

We hope this Deep Dive gave you some useful frameworks for thinking about it.

Keep exploring.

Keep questioning how these concepts affect your digital life.

Thank you for joining us.

A warm thank you from the Deep Dive team.