What is edge computing?

It is great to have you here! Before we dive into the world of edge computing, I would love to learn a little bit about your background so we can tailor this session perfectly to you.

To help me set the right pace, what is your current understanding of cloud computing, and what is your main reason for wanting to learn about edge computing today?

I know the basics of cloud computing and want to understand how edge computing is different.

Flipping the Cloud Inside Out

To understand edge computing, it helps to look at how we got here. Traditional acts like a massive, highly centralized brain. When you use a cloud service, your local device behaves almost like a dumb terminal: it gathers information, packages it up, and sends it on a long journey across the internet to a distant data center. That central data center does all the heavy lifting—the processing, sorting, and analyzing—and then beams the results all the way back to you. While this centralized setup is incredibly powerful for storing files and running massive simulations, it creates a physical bottleneck. If your data has to travel thousands of miles round-trip just to get a basic response, you are going to experience delay. This lag is what engineers call , and in our fast-paced world, even a fraction of a second can feel like an eternity. Edge computing completely flips this model on its head. Instead of forcing all data to make that long pilgrimage to a centralized data center, edge computing brings the processing power directly to the "edge" of the network—physically closer to where the data is actually generated. Think of it as distributing little helper brains closer to the action, whether that is on your smartphone, a local smart thermostat, or a nearby micro-server.

A side-by-side architectural comparison showing centralized cloud computing on the left versus distributed edge computing on the right.

I have set up a quick visual to help you see this architectural shift in action. Notice how data packets in the edge model have a much shorter distance to travel, which keeps communication snappy.

Now that you see the basic layout, let's explore what actually lives at the "edge" of a network.

What are some everyday examples of 'edge' devices?

That is a great question because "the edge" can sound abstract, but you probably interact with edge devices every single day without realizing it. Broadly speaking, an is any piece of physical hardware that has enough built-in processing power and memory to analyze data and make decisions right on the spot, rather than waiting for a distant cloud server. This includes everything from the smartphone in your pocket to and advanced industrial sensors.

Everyday Examples of Edge Tech

Let's look at three common scenarios where local processing makes a massive difference:

Your Smartphone's Face Unlock: When you glance at your phone to unlock it, the facial recognition math happens entirely on a specialized chip inside your phone. If your phone had to send your raw facial scan to a cloud server $1,000 miles away just to authorize you, the delay would be incredibly frustrating. More importantly, keeping that biometric data locked on your personal device is far more secure.
Smart Traffic Cameras: Modern traffic cameras do not just record video; they use built-in edge processors to count vehicles, spot accidents, and detect speed violations on the fly. Instead of streaming gigabytes of heavy, raw video footage back to a central city database 24/7, the camera processes the video locally and only sends back tiny text alerts when something notable happens, saving immense network bandwidth.
Wearable Health Monitors: Smartwatches that monitor your heart rate can detect anomalies like irregular heart rhythms in real time. Because the watch itself processes the raw sensor data, it can vibrate and alert you of an emergency instantly, even if you are hiking in an area with absolutely no cellular service.

In all these cases, processing data right where it is collected keeps operations fast, reliable, and private.

what about cdns? in relation to a high traffic website like pinterest.

Caching at the Edge: CDNs

You hit on a brilliant connection! A is actually one of the earliest and most widespread forms of edge computing, even if we did not always call it that.

To see how this works, let us look at Pinterest. Imagine Pinterest has one main cloud database in Virginia where it stores every single pin, image, and user profile. If you are browsing from Tokyo, and your phone has to request a high-resolution image of a living room remodel all the way from Virginia, that data has to travel across undersea cables, through dozens of routers, and back. That takes time.

Instead of making that long trip for every single user, Pinterest uses a CDN. A CDN is a massive network of servers scattered all over the globe, located in local cities. The CDN server in Tokyo acts like a local storage vault. The first time a user in Japan requests that living room pin, the local Tokyo server grabs it from Virginia and keeps a copy—a process called caching. When the next thousand users in Tokyo want to see that same pin, the CDN server delivers it to them instantly from just a few miles away.

A schematic mapping of data request paths from a user in Tokyo to a central server in Virginia versus a local edge server.

The diagram shows how a CDN intercepts requests and serves them locally, keeping load times incredibly fast.

But here is where the transition to modern edge computing gets exciting. Traditional CDNs only served "static" files—things that do not change, like images or video clips. Today, modern CDNs can actually run code at those local servers.

For example, if Pinterest wants to customize your feed based on your device type or translate a button into Japanese on the fly, it can run that minor bit of computer logic directly on the Tokyo CDN server. This is called edge scripting, and it means the response is not just fast—it is customized without ever involving the central Virginia database.

How does a CDN decide when to update its cached files?

That is the million-dollar question! If a CDN kept copies of Pinterest's images forever, you would quickly run into two massive problems: the local servers would run out of storage space, and users would eventually see outdated content.

To keep everything fresh and efficient, CDNs rely on two core concepts: TTL (Time-to-Live) and cache invalidation.

The Expiration Date: Time-to-Live

Think of every cached file as having a built-in expiration date stamped on it. In the networking world, this stamp is called .

When Pinterest's main server in Virginia sends an image to the local Tokyo CDN server, it attaches a header specifying how long that image should be considered "fresh"—for example, 24 hours. As long as those 24 hours haven't ticked away, the Tokyo server will happily serve that cached image to every local user instantly.

Once the TTL clock runs out, the cached file is marked as "stale." The next time a Tokyo user requests that pin, the Tokyo server doesn't just hand it over. Instead, it takes a quick trip back to the Virginia server to ask: "Has this image changed?" If it hasn't, Virginia tells Tokyo to renew the lease, reset the TTL clock, and keep using the cached version. If it has changed, Tokyo pulls down the brand-new version, caches it, and starts a fresh countdown.

The Panic Button: Cache Invalidation

But what happens if a Pinterest user deletes a pin or updates their profile picture, and you can't wait 24 hours for the TTL to expire? You don't want other users seeing a deleted image for a whole day.

This is where comes in. This is the manual

When an update occurs, Pinterest's central server sends an urgent, global "purge" message to the CDN. The Tokyo server receives this alert, immediately throws away its cached copy, and is forced to fetch the updated file from the central cloud database the very next time a user asks for it.