Skip to content Skip to footer

IP Anycast: the technology that makes the internet work as fast as you want it to

The internet has become an indispensable part of our lives, and the speed with which we access information and services online is crucial. Behind this seemingly instantaneous experience is a technology called IP Anycast. As such, this technology plays a crucial role in optimizing the speed and increasing the efficiency of the internet.

By allowing Internet traffic to be routed to the nearest available server, IP Anycast guarantees a faster and more reliable browsing experience for users, as well as providing greater resilience and scalability for service providers. In this article, we’ll explore how IP Anycast works and its impact on the way the internet works, offering an in-depth look at this fundamental technology. Happy reading!

Putting IP Anycast into context

To better understand IP anycast, we need to put it into context. In one of his latest books, entitled “Click Here to Kill Everyone“, cryptologist Bruce Schneier, considered one of the world’s most experienced cybersecurity experts, observes that, in general, everything is connected. So he realizes that we are all connected.

For the author, we have reached such a degree of dependence and interdependence on networks and the Internet that, when they become systemic, certain failures could be disastrous. Despite this risk, we want and need networks. Because they have brought us comfort, security, leisure, health and many other advantages. Want to see one? Pay-TV services, for example. Another one? Internet banking and mobile banking services. Not to mention messaging apps, video apps, social networking apps, health apps and many others.

But we didn’t reach this level of interconnectedness overnight. This was a constructive process. Over the last thirty years, network technology has been challenged by Internet users to deliver more and more content. In addition, it was challenged to deliver with ever-increasing data density and speed. The challenges were and still are many, but they are being overcome one after the other. The main key to this was the progressive increase in connection speeds. In Brazil, we can currently enjoy domestic broadband with a nominal rate of 600 Mbps or more. However, when the Internet arrived in the country in May 1995, home users used modems that connected to dial-up lines at speeds that rarely exceeded 28.8 Kbps.

More speed: an evolution demanded by users

However, little by little, the combination of supply and demand for faster connections led to advances in network technology. In addition, speeds have increased and, as a consequence, services have appeared that previously could not exist. One example of this is video streams on pay-TV services.

So everything we see today is built on network technologies that weren’t available thirty years ago. They are responsible for the magic that allows you to participate in a videoconference with people in different parts of Brazil or abroad. Thirty years ago, you would suffer from delays in the display of pages and the occasional loss of connection due to network or server overload.

The solution to this type of problem was a combination of several of these network technologies, one of the most notable being packet switching. This expression refers to the grouping of data into “packages” made up of two main parts. The first is a header with the addressing data for the delivery of this “packet”, and the second is a “payload”. In other words, it works just like an envelope you send to someone with a document inside. In this example, the post office is the network, which will use the address on the envelope (the header) to transport it to the recipient. It’s up to the recipient to open the envelope. This is where IP Anycast comes into its own.

How data travels on networks

In data communication networks, this is precisely the simplest type of connection. It is called “unicast” and allows a device to be connected to only one other device. This way, communication will only take place between these two – even if there are other devices on the same network.

Because of their nature and because of the needs of applications and users, however, networks offer different “delivery services”. With broadcasting, they allow one device to send a message to all the others on the same network. With multicast, they address messages (actually data packets) to groups of devices.

All of this is reasonably simple when it comes to a local network, in which the devices are all in the same house, the same building, the same store. But when the network we are discussing is distributed over a large geographical area (for example a WAN or wide area network), other problems arise and need to be solved. And if we’re talking about the Internet, then the problems are all the greater.

So there are several problems, each with its own solution, but one of the most important and annoying is latency. If you’ve ever pressed a button to call an elevator, you know exactly what latency is. It’s the time it takes for the elevator to arrive. On a network, it’s the time it takes for the computer on the other end to respond. But you also know the condition in which the elevator arrives fastest: it’s when it’s very close, just one floor above or below yours.

Content far from the user is always a problem

The big content providers, such as video streaming, social networks and news, generally generate this content at a certain point on the planet, but they need to deliver it to users all over the world. If the generation is in the city of São Paulo, great for those who live there: because between the router in the user’s home and that content producer’s network there are probably five or six other networks along the way. Thus, the data packet’s journey will only take five or six steps – five or six “hops” (jumps between networks).

But everything changes if the content is on the other side of the world, with 20, 25, 30 jumps or more. Without the use of good network technology, waiting for this content is like calling an elevator in a commercial building at six in the evening: it may be crowded several times before you can get off.

Network technology solved this problem just as engineers solved the problem of overcrowded elevators in large buildings. The solution has two components, one hardware and one software. The hardware component is the sufficient number of elevators, calculated during design and installed during construction; the software component is the elevator management system – when it’s time to leave, you just press a button (or enter zero on your floor, indicating that you want to go to the first floor) and the system sends the elevator that’s closest to your floor.

Closer data arrives faster

In networks, this method is called “IP anycast”: you click on a link in a social network, and you’ll receive the data from the server closest to your home. To make IP anycast work, content providers – or the companies that provide them with network infrastructure, such as Huge Networks – install (or hire) servers at strategic geographical points. These servers – all with the same IP address – store copies of the content to be distributed. This way, when they receive a request for content from a user, the request will be fulfilled by the server closest to them. Therefore, the server to which the request arrived with the fewest hops.

Because of these characteristics, IP anycast is the basis of content distribution networks. It favors data delivery with the shortest possible latency time. But that’s also why anycast is a defense tool for servers against denial of service attacks. During these attacks, cybercriminals launch millions of service requests to a given server, causing overload and service interruption. Anycast has several features for defense, starting with the possibility of distributing requests across the available servers (all with the same IP, remember?), just like Huge Networks’ HugeGuard Cloud service does.

Arquitetura IP Anycast

IP Anycast and protection for customers on networks

Other features available in “anycast” to protect customers on the networks:

1) Traffic shaping: controlling traffic to ensure that critical applications receive the most bandwidth;

2) Redirection: attacker requests are identified and redirected to specific network nodes, thus cleaning up and avoiding server overload;

3) Rate limiting: limit on the rate at which incoming traffic is processed on a server. If a DDoS attack is identified, the rate can be increased to allow legitimate traffic to be processed while malicious traffic is filtered out;

4) Traffic filtering: anycast can identify traffic patterns from specific regions (identified by IPs) and block them in the event of suspicious activity. This approach ensures that malicious traffic never reaches the network.

Go to Top