DDoS protection: Appliance, ISP or cloud? What actually protects you when it counts
Appliance, ISP or cloud: Which DDoS protection holds up in an emergency? A comparison with practical insights from logistics company GLS.

There are three effective ways for companies to protect themselves against DDoS attacks. They can run a hardware appliance in their data center, rely on their internet service provider for protection, or engage a specialized cloud provider. Germany's Federal Office for Information Security (BSI) distinguishes between appliances, content delivery networks (CDNs), and DDoS mitigation as a service in its criteria for DDoS mitigation. All of these approaches filter attack traffic. The decisive difference, however, is where they do it.
Christof Weppler knows what this question looks like in practice. He is a senior IT architect at the parcel service provider GLS and experienced a severe DDoS attack firsthand. He shared his experience on the Link11 podcast Follow the White Rabbit. His insights are featured in this article.
Key takeaways
- An on-premises appliance protects your services but not your internet connection.
- ISP-based protection sits in front of the line but ties you to a single provider.
- Cloud-based DDoS protection filters out attacks before they reach your infrastructure and is independent of your ISP.
- Network protection on Layers 3 and 4 forms the foundation. Applications and APIs also require Layer 7 protection.
How do DDoS attacks take down a network?
In a Distributed Denial of Service (DDoS) attack, a target is flooded with traffic from many sources. Most often, the internet lines themselves are affected. Once the bandwidth is exhausted, legitimate traffic cannot get through.
This is exactly what happened at GLS. At the time, the company was connected via two lines, one with 1 Gbit/s and the other with 500 Mbit/s. An attack saturated both inbound lines at 100 percent. The web portal went offline. Handheld scanners in the depots stopped working, causing sorting to come to a standstill, and preventing drivers from delivering parcels. Even a few minutes of downtime can cause significant damage in logistics.
The threat landscape has become noticeably more severe. According to the Link11 European Cyber Report 2026, the number of attacks on the Link11 network increased by 75 percent in 2025. The longest single attack lasted 12,388 minutes—more than eight days. Although the number of attacks in the first half of 2026 fell by 42 percent, their intensity reached record levels. Additionally, there are tactics like carpet bombing, in which attackers hit entire address ranges simultaneously.
DDoS protection compared: three approaches and their limits
On-premises appliance: protection behind the bottleneck
A DDoS appliance is a device located directly within your network. It inspects incoming traffic and filters out attacks before they reach your servers and applications. Advantages include full control and relatively fast deployment.
However, its position is a weakness. The appliance sits behind the internet line. If a volumetric attack overloads that line, not even the best filter can help. For Christof Weppler, this option was quickly ruled out: "If the path into the data center is blocked, nothing works."
His attempts to mitigate the attack confirmed this. The team blocked the attack traffic at the BGP router. BGP (Border Gateway Protocol) controls the paths that traffic takes into a network. This allowed the firewall, which had a full session table of nearly one million entries, to recover. However, the lines remained saturated.
ISP-based protection: a first step
Many companies rely on their internet service provider (ISP). ISPs filter attack traffic within their own networks, before it reaches their customers. This is precisely what happened at GLS: one of the providers swiftly implemented a filter list and blocked the attack traffic.
GLS then protected one of its two lines through the provider. During an attack, the team rerouted all traffic via BGP over this line. This approach worked reliably for years.
However, this approach has its limits. The protection is tied to a single provider, and each additional line remains a weak spot. Added to this are limited filter capacity, often only coarse filters, and unclear SLAs and response times. When GLS planned to relocate its data centers, this dependency became an obstacle. The existing solution had not failed, but it depended on the infrastructure of a single provider. GLS therefore looked for provider-independent protection and, after a market analysis, chose Link11.
Cloud-based DDoS protection: independent and in front of the line
Specialized providers filter out attack traffic within their globally distributed infrastructure. Only clean traffic reaches your lines. Your bandwidth remains free regardless of the size of the attack or your ISP.
GLS has chosen this approach. Christof Weppler and the IT Security team conducted a market analysis and compared three vendors based on the scope of their services, support, price, and company size.
The choice was Link11. The partnership was a deciding factor besides the technology. "With Link11, we are not a small fish. We communicate as equals," says Weppler. The BSI's list of qualified DDoS mitigation providers also helped with the selection process.
Where does network protection end and where does application protection begin?
Network-based DDoS protection covers layers 3 and 4, or the network and transport layers. These are the layers where volumetric and protocol-based attacks operate. It forms the foundation on which everything else is built.
However, many attacks target the application layer (Layer 7). These attacks target web portals, login pages, and APIs. According to the latest Thales report, bots now generate more than half of all web traffic. Whether a request is legitimate can only be determined by understanding the specific application.
GLS is aware of this as well. Numerous bots automatically query parcel tracking status there. The team limits this access using rate limiting; anyone who sends more than ten requests per minute is placed on a blocklist for five minutes. Christof Weppler draws a clear line. The network team is responsible for layers 3 and 4, while the application layer is the responsibility of the owners of the portals and APIs, together with IT security.
Planning DDoS protection: which questions should you answer first?
Before talking to vendors, take a closer look at your infrastructure. Christof Weppler suggests asking yourself three questions: Where are your weak spots, and through which paths does traffic enter your network? Which processes depend on your critical paths? Which attacks have you already experienced, and did they take place more on the network or the application layer?
Think through every attack to the end. What happens if a line becomes completely saturated? How quickly can you reroute? At GLS, the chain extended from the data center to the driver at the customer's door.
Conclusion: What matters is where DDoS protection kicks in
An appliance protects your services but not your lines. Your ISP protects your lines but ties you to its network. A specialized cloud provider, on the other hand, filters out attacks before they reach your infrastructure while remaining independent of your ISP. For companies whose business depends on constant availability, this is the most resilient solution.
Link11's Network DDoS Protection filters attacks in the cloud, keeping your bandwidth free. Your services stay available, even when it counts. Listen to the full conversation with Christof Weppler on the Link11 podcast Follow the White Rabbit (episode available in German only).
DDoS protection that explains what it mitigates.
Network DDoS Protection filters in front of the uplink, learns your normal behavior and logs every intervention.



