Network security
Understanding Layer 3/4 Volumetric DDoS Attacks vs. Layer 7 Application Attacks
Distinguish bandwidth floods, connection exhaustion and costly HTTP requests, then match each threat to the right defensive layer.
Identify the resource being exhausted
A distributed denial-of-service attack uses traffic from multiple sources to make a service unavailable. The limiting resource may be bandwidth, packet-processing capacity, connection state or application work. Looking only at gigabits per second can miss an attack that consumes database or PHP capacity with relatively little traffic.
Layers 3 and 4 refer to network and transport behaviour; Layer 7 refers to the application. A SYN flood targets TCP connection handling and is not necessarily a bandwidth-saturating attack. Separate packet volume from connection-state exhaustion when choosing controls.
Different attacks need different evidence
Compare attack signals with the normal baseline for each service. Legitimate events can also produce large spikes, so a high rate alone is not enough to classify traffic.
| Pattern | Likely pressure | Relevant controls |
|---|---|---|
| UDP/reflection flood | Link bandwidth or packet processing | Upstream filtering and traffic scrubbing |
| TCP SYN flood | Connection state or packet processing | SYN protection, connection controls and network mitigation |
| HTTP flood | Web workers, application logic or database | WAF, behavioural controls, request limits and safe caching |
Filter before the constrained link
If attack traffic fills the access link, a firewall behind that link cannot recover the lost capacity. Network mitigation needs to happen upstream of the bottleneck. Scrubbing systems classify and filter traffic, while routing can direct traffic toward the mitigation path.
BGP is part of traffic steering and route policy; it does not inspect HTTP requests or distinguish a costly search from a normal page view. Blackholing can contain an attack by dropping traffic to a target, at the cost of that target’s availability. Agree escalation and activation procedures before an incident.
Protect expensive application paths
An HTTP flood can use valid requests against login, search or uncached endpoints. Application protection needs request-level visibility, normally at the point where HTTPS is terminated. Use WAF rules, endpoint-specific rate limits, bot controls and caching where responses can be safely shared.
Test limits against shared office networks, mobile carrier NAT, legitimate APIs and monitoring. Blocking every busy source can deny real users. Observe application latency and database load alongside blocked requests, and keep a way to roll back an overly broad rule.
Combine network and application protection
Gloman’s cybersecurity service combines DDoS scrubbing at the AS50311 routing edge with an edge WAF and bot mitigation. These are complementary controls: the network layer addresses traffic and connection pressure, while application controls address HTTP behaviour.
Confirm protected IPs, protocols, TLS arrangements, capacity scope and response responsibilities for your service. Keep origins reachable only through authorised paths where possible, and include DNS and management access in the incident plan. No single mechanism guarantees that every attack is neutralised.
- Record normal bandwidth, packets per second, connection rates, HTTP request rates and application latency.
- Define who can change routes, activate mitigation and approve disruptive containment.
- Rehearse communication and recovery using agreed, controlled tests.
- After an incident, review legitimate-user impact as well as attack volume and tune the controls accordingly.