Fail-safe Internet: How does redundancy actually work?

Especially in the event sector, where everything is live, fail-safe Internet plays a paramount role.

You've probably heard this tip before when you have Internet problems: "Just restart the device". At home, this is no problem and actually helps in many cases. But during a live production? There, of course, it's completely unthinkable. But when it comes to Internet connections, the question of bandwidth, i.e. speed, is usually at the forefront. But there is one point that is just as important: the reliability of the connection.

But there is no such thing as a truly fail-safe Internet. An Internet connection failure can happen at any time and for any reason. In today's digitized world, this quickly means the loss of livestreams, cloud-based services or complete communication. You should therefore always be prepared for a connection failure - at any time. Fortunately, there are solutions to greatly reduce the risk.

The word "redundancy" usually comes up in this context. But what does that mean in concrete terms? And where are the differences in the individual approaches? That's what this article is about.

What is redundancy?

Redundancy initially means only the provision of technically identical systems in multiple versions. In normal operation, the redundant components are actually superfluous. But as soon as the extreme case occurs and a system fails, the redundant system comes into play and takes over the respective task. With fail-safe Internet, for example, this means multiple connections instead of just one. Wikipedia explains this in a bit more detail.

However, this still says nothing about how the handover from a failed system to the replacement system takes place. Does an employee simply switch the cable by hand? Should the switchover to the replacement line happen automatically? And how quickly does this have to happen?

Solutions for redundant connections

A normal Internet connection is not a redundant connection. If the connection fails, the signal is gone and you can only wait and hope that it will work again quickly. This is not a good option for a digital infrastructure at events or event locations.

It is better to keep one or more backup connections on hand - depending on how crucial a continuous Internet connection is for the project. The redundant connections can be a second DSL connection, a satellite uplink or an LTE router, for example.

There are various methods for this. We would like to present the solutions backup, failover and bonding here.

Backup

A backup connection is simply a second available connection that can be used in an emergency to make the Internet fail-safe. If the signal of the main connection fails, the cable is simply reconnected. This method is easy to implement, since only a second connection needs to be provided. However, someone must always be on standby to recognize that there is a problem and then quickly put the second connection into operation.

Since the replacement connection dials into the Internet with a different IP address, all services that require a login are interrupted. For example, participants in live streams have to dial in again, or users of online services such as mail services or cloud storage have to log in again. It is usually not noticeable when just surfing the Internet.

Duration of the interruption	New login required Specifies whether users need to log in to services (such as livestreams or online accounts) again after the connection is restored.	Recommended product
Several minutes	Yes	Hotspot in a box

Failover / Same-IP failover

A failover solution automatically switches to the redundant connection in the event of an interruption. To do this, you have to set when the main line is considered to be interrupted. In most cases, test packets, so-called pings, are sent to a server on the Internet that is considered reliable. If the server confirms receipt, the connection is considered active. However, this carries the risk that the connection will be classified as "interrupted" even if the server in question does not respond for other reasons. For example, because it has a problem itself. It is therefore recommended to always send the pings to at least two independent servers. A common setting is to classify the line as dead after three unanswered pings from two servers and to initiate the switch to the replacement line.

The replacement connection also dials into the Internet with a different IP address during failover. As with the backup, this may cause services like livestreaming or online services to require users to log in again.

As a solution, we offer "Same-IP Failover". Here, each signal is first sent (via VPN tunnel) to our data center and from there routed to the Internet with the always same IP address. This means that the connection "feels" the same for all applications on the Internet, and even security-critical services do not require a new login. Even if the actual connection fails and the user therefore switches to the substitute connection.

Duration of the interruption	New login required Specifies whether users need to log in to services (such as livestreams or online accounts) again after the connection is restored.	Recommended product
10 - 20 seconds	Single failover: Yes For Same-IP failover: no	Multi LTE Case

Bonding

With signal bonding, all connections used are sent (via VPN tunnel) to a data center, where they are converted into a bundled signal. From there, the signal always goes to the Internet with the same IP address. The available connections are always active during bonding and can be used in parallel. All data packets are sent simultaneously over all lines. The packet that arrives first at the data center is accepted. All others are discarded. We have described the exact function of bonding in another blog article.

The advantage of bonding is that there are no switchover times if one of the connections used fails. This is because all connections run in parallel throughout. This is especially useful for live streams, since the interruption of one line is not noticed at all. With two independent connections, the setup can already be called quite fail-safe Internet. If you use two or more instead of just one spare connection, the connection can be operated redundantly several times. The risk of failure is virtually zero. However, this method always requires a remote station in the data center where the signals are synchronized and bundled. As with the other solutions, you can rent the bonding, including three LTE connections, from us on a project-by-project basis. In conjunction with a house connection, you thus already have fourfold fail-safety.

Duration of the interruption	New login required Specifies whether users need to log in to services (such as livestreams or online accounts) again after the connection is restored.	Recommended product
0 seconds (no noticeable failure)	No	Multi LTE Case

 

The right solution for fail-safe Internet ultimately depends on the project in question. Some applications, such as terminals for cashless payments, can cope with a few minutes of connection failure. For livestreams, on the other hand, it's the super-GAU if the connection goes down for even a second.

We will be happy to advise you on your project.