DDoS Attack Types and an Enterprise Protection Strategy
Nanda Teknoloji Engineering Team · · 7 min read
In brief
DDoS attacks fall into three main types, volumetric, protocol, and application-layer, and each requires a different defense. An effective strategy combines traffic filtering at the network edge, WAF and rate limiting at the application layer, a pre-prepared response plan, and regular drills.
A DDoS (Distributed Denial of Service) attack is an attempt to exhaust a service’s capacity with traffic from many sources and block legitimate users’ access. Attacks fall into three main types: volumetric attacks that saturate network bandwidth, protocol attacks that exhaust connection tables and protocol resources, and application-layer attacks that target the application itself. Each type creates a bottleneck at a different point, so defense is built not with a single product but with a layered strategy. A sound plan includes traffic filtering at the network edge, a WAF and rate limiting in front of the application, a pre-written response procedure, and regular drills. Below we cover each type, the right defense, and the preparation steps.
Why are DDoS attacks a business problem?
DDoS is not a data breach; its target is availability. But the business impact is large: a store that goes down on campaign day, a customer portal that cannot be reached, or an API that times out. Some attacks are also used as a distraction; while the security team deals with traffic, another attempt may be underway. For this reason, DDoS preparedness is not only the network team’s concern but part of the business continuity plan.
DDoS attack types
Volumetric attacks
The aim is to fill the target’s internet connection or upstream bandwidth. With reflection and amplification techniques (such as abusing some UDP-based protocols), the attacker turns a small request into a large volume of response traffic. Once the line is full, no request can get through even if your server is sound. This attack cannot be filtered on your side, because the line is already clogged before the traffic reaches your organization’s door.
Protocol attacks
These exploit weaknesses in the network and transport layers. SYN flood is the best-known example: the attacker starts the connection handshake but does not complete it, and the server’s connection table fills with half-open connections. Firewalls, load balancers, and the server operating system have limited resources in these tables; the target is often not bandwidth but connection capacity.
Application-layer attacks
These are carried out at the HTTP/HTTPS level with requests that look legitimate. For example, sending an expensive search query thousands of times per second, triggering a heavy report endpoint, or continuously refreshing the login page. The traffic volume is low, but every request tires out the database and the application server. These attacks are the hardest to distinguish, because the individual requests look normal.
| Type | Target resource | Typical symptom | Main defense |
|---|---|---|---|
| Volumetric | Bandwidth | Line saturation, unreachability across all services | Traffic filtering at the network edge, distributed mitigation |
| Protocol | Connection tables, firewall | High half-open connection count, device CPU load | SYN protection, protocol validation |
| Application layer | Application and database | Low bandwidth but high CPU and latency | WAF, rate limiting, bot challenges, caching |
Protection strategy: a layered approach
A single device or rule does not cover every type. The layers below complement each other.
- Edge mitigation. Volumetric traffic must be filtered in a high-capacity distributed infrastructure before it reaches your organization’s line. This is not a problem that can be solved by increasing your own line capacity.
- Distributing and caching traffic. A CDN serves static content at the edge, reducing the load on the origin server and spreading attack traffic across a broad surface. For details, see our CDN article.
- Application-layer protection. When WAF and DDoS protection work together, application attacks are filtered out through request rate, client behavior, and signature rules. For how a WAF works, see the what is a WAF article.
- Hiding the origin server. If the back-end IP address is known, an attacker can bypass the protection and attack the server directly. Open the back end only to traffic coming from the protection layer.
- Architectural resilience. Automatic scaling, load balancing, and decoupling dependencies (database, third-party APIs) limit the impact of an attack.
Response plan: what to write before the attack
If decisions are made during the attack, time is lost. The following should be settled in advance:
- Normal traffic baseline. “Abnormal” cannot be defined without knowing hourly request counts, country distribution, and typical response times.
- Roles and communication. Who decides, who talks to the protection provider, and who informs business units and customers?
- Thresholds and automatic rules. At which request rate does which measure kick in automatically?
- Critical service priority. If capacity shrinks, will the payment flow or the public storefront be protected first?
- Fallback options. Are a maintenance page, a static backup page, and alternative routing ready?
- Post-attack review. Logs are preserved, and the root cause and an improvement list are written.
A scenario: the quiet attack on campaign morning
On campaign morning, an e-commerce company sees response times increasing. Network traffic is at normal levels, but database CPU has risen. Investigation shows that a large number of requests, each containing a different query, are hitting the product search endpoint; this is an application-layer attack. Volumetric protection shows nothing here, because bandwidth has not filled up. The solution is to add rate limiting on the search endpoint, a bot challenge for abnormal client behavior, and short-lived caching for popular queries. We cover similar peak-period preparations separately in our article on campaign-period infrastructure preparation.
Common mistakes
- Activating protection after the attack has started. Traffic rerouting (a DNS or routing change) may need propagation time.
- Preparing only for volumetric attacks. Low-volume application attacks are often overlooked.
- Leaving the origin IP exposed. Old DNS records and email headers can leak the address.
- Not running drills. If the plan stays on paper, it gets tested for the first time in a real attack.
- Not caching cacheable content. Having every request go to the origin enlarges the attack surface.
- Neglecting communication. If customers and business units are not informed, reputational damage occurs even when the technical fix is fast.
How to proceed during an attack
Following an orderly flow instead of panicking lowers the margin for error during an attack.
- Confirm. Is the slowness really an attack, or a deployment error or a legitimate traffic increase? A recently released version and marketing notifications should be checked.
- Identify the type. If bandwidth is filling up, it is a volumetric attack; if connection tables are filling up, a protocol attack; if only application resources are being tired out, an application-layer attack.
- Engage the protection provider. If traffic filtering or rerouting is needed, notify the provider through the pre-defined channel.
- Narrow the impact. Tighten rate limits on non-critical endpoints, temporarily turn off expensive features, and, if necessary, apply additional challenges for certain geographies.
- Run communications. Give internal stakeholders and, if necessary, customers a plain status message; avoid making promises about the estimated resolution time.
- Preserve the evidence. Keep logs, traffic samples, and the timeline; they are needed for the post-incident review and, if necessary, for legal notifications.
Reviewing thresholds and rules after the attack ends is essential. The same attacker usually returns with a different method; the output of each incident is the input of the next defense.
Readiness checklist
- Has a current inventory of all externally exposed domain names and IP ranges been made?
- Is DNS management set up so that quick rerouting with a low TTL is possible?
- Are the escalation channel and responsible people with the protection provider written down?
- Is there rate limiting for expensive endpoints in the application (search, report, export)?
- Can a backup “maintenance” or static page be put live quickly?
- Are the dates of the last two drills and the resulting actions recorded?
This list also forms the skeleton of a short readiness report to be presented to the board. To measure the level of readiness, it is a good habit to review the checks at least once a year, and separately after major architectural changes.
Decision summary
| Situation | Priority step |
|---|---|
| Your line is limited, and you depend on a single provider | Edge mitigation service |
| You see application attacks and bot traffic | WAF, rate limiting, bot management |
| High share of static content | Move the load to the edge with a CDN |
| Plan and roles are not written down | Response procedure and drills |
For a general framework, the NIST Cybersecurity Framework document on cybersecurity risk management helps you think about defensive capability under the identify-protect-detect-respond-recover functions.
Moving forward with Nanda Teknoloji
Nanda Teknoloji carries out the discovery, design, migration, and operations steps in projects like this: mapping the current attack surface and traffic baseline, designing layered protection, writing the response plan, and operating it. For a structure that fits your organization’s surface, you can review our WAF and DDoS protection solution.