Response time is one of the key metrics used to evaluate the effectiveness of DDoS protection. It depends on several factors, including how quickly an attack is detected and classified, how long it takes to apply filtering rules, and how effectively those rules mitigate malicious traffic. But what exactly does response time measure, and what guarantees should you expect from your provider under your service agreement? This article breaks it down.
DDoS incidents are becoming longer-lasting. According to our Q2 2026 data, the average duration of network attacks targeting financial companies increased by 452% year over year, while the most powerful attacks against telecom and hosting providers lasted for weeks. That makes it important to understand when choosing an anti-DDoS service how long it can take from the start of an incident until all malicious traffic is filtered out.

Key Takeaways
Responding to a DDoS attack involves three main stages, each with its own starting point:
- Detection: when a threshold configured in the Client Portal is exceeded.
- Support response: response times vary by service plan. At StormWall, they range from 15 minutes to 3 hours, depending on the selected support level.
- Mitigation: the time required depends on the complexity of the attack.
The actual response time for a serious incident depends on the specifics of the attack. You can review how the service handled a particular attack using the data available in the Client Portal.
What Does DDoS Response Time Actually Mean?
DDoS mitigation typically involves several scenarios. They may occur independently of one another:
- A filtering rule is already in place and is triggered as soon as an attack begins, automatically filtering out malicious traffic.
- The service becomes unavailable, and the customer contacts the anti-DDoS provider’s support team. Engineers then create additional filtering rules.
- An engineer monitoring traffic detects the attack first. They adjust the rules to account for the characteristics of the new incident, making filtering more effective.
| Metric | What it measures | Starting point | Where to find the data |
| TTD (Time to Detect) | The time from the start of an attack to the first alert. It shows how quickly monitoring systems detect an attack. | The actual start of the attack, marked by the rise in traffic on the graph. | The interval between the point at which traffic begins to rise on the graph and the timestamp shown in the Start field of the attack report or notification. |
| TTR (Time to Respond) | The time it takes for an engineer to respond. | When the support request is registered. | The support ticket or chat message. |
| TTM (Time to Mitigate) | The time it takes to mitigate the attack. | The moment the attack is detected (the Start field in the report). In on-demand mode, the clock starts when traffic is redirected for scrubbing. | On the traffic graph, measure the time from the timestamp in the Start field until traffic reaching the server returns to normal levels, even if the attack is still ongoing. |
These metrics can range from seconds to hours. An agreement with an anti-DDoS provider often specifies only the second metric. The first depends on your configuration, while the third is affected by the provider’s support team and the attack’s scale and complexity.
When discussing response speed, it’s important to clarify which event is being measured and which service plan applies. Let’s look at each metric in more detail.
1. Attack Detection Time
For monitoring purposes, an attack is detected when a traffic threshold is exceeded and/or one or more detection rules are triggered. You can see which rules were triggered in the Reason field of the attack report.
How quickly an attack is detected and filtering begins depends on the protection mode:
- As little as 1 second in always-on mode, where traffic is continuously scrubbed and filtering rules are applied immediately.
- As little as 30 seconds in on-demand mode, where filtering is activated when needed. There are two possible scenarios: the customer can manually announce IP prefixes, in which case response time depends on the customer’s actions; or a DDoS sensor can trigger protection automatically, usually within 30–60 seconds after the threshold is exceeded.
Protection settings are configured individually and can use different indicators to identify the start of an attack. For example:
- For networks: the increase in packet rate.
- For websites: the share of error responses and whether the RPS threshold has been exceeded.
Most of our customer traffic monitoring is automated. In addition, our engineers work around the clock, tracking a wide range of metrics with specialized tools.
2. Support Response Time
Support response times vary from one anti-DDoS provider to another and may also differ across service plans offered by the same provider. Here’s how response times compare across StormWall plans:
| Product | Plan | Support response time | SLA availability guarantee |
| StormWall for Web, StormWall for Networks | Personal | 3 hours | At least 98% (StormWall for Web) and 99% (StormWall for Networks) |
| Business | 60 minutes | At least 99.2% | |
| Enterprise | 15 minutes | At least 99.5% | |
| StormWall for Servers | Standard | Up to 60 minutes | 99.2% |
| Business | Up to 30 minutes | 99.5% | |
| Enterprise | Up to 15 minutes | 99.9% |
How Support Tiers Work
Anti-DDoS providers often organize their technical support teams into several tiers, each with its own responsibilities:
- Tier 1 receives and registers support requests, then determines where to route them — to Tier 2 or directly to a specialized team, such as network engineers.
- Tier 2 handles routine issues.
- Tier 3 consists of engineers with specialized expertise and in-depth product knowledge. They handle the most complex issues and identify the best way to resolve unusual or non-standard problems.
Support response times remain the same at night, on weekends, and on public holidays: specialists are available around the clock. The guaranteed response time is determined by the service plan, not the time of day.
The situation can be different on the customer’s side, however. Their own specialists may not be available outside business hours. For example, at midnight, there may be no one available to confirm that a traffic spike is caused by a sale rather than an attack.
3. Attack Mitigation: Rules and Iterations
There is no single starting point for measuring mitigation time because traffic filtering relies on a cascade of rules. Each packet passes through the first rule, then the second, the third, and so on, until a rule blocks it or allows it through. If none of the rules match, the packet passes all the checks and reaches the server. This process takes fractions of a millisecond: packets are inspected on the fly as traffic flows through the system.
When an attack is complex or involves multiple vectors, engineers add or refine filtering rules. Each iteration blocks another portion of malicious traffic, so mitigation may take place in several stages.
There are two main types of rules:
- Unconditional rules take effect immediately. They inspect every packet as soon as traffic enters the scrubbing network. For example, if a rule blocks packets with a specific header, any matching packet is dropped immediately.
- Threshold-based rules take effect only when a specified condition is met. For example, a rule might block traffic from an IP address after it sends 100 packets within 10 seconds. If an attacker sends 99 packets during the first second and the 100th packet at the 10-second mark, the traffic will only be blocked at that point. This is why threshold settings also affect attack detection time.
Protection methods can be divided into two categories:
- Non-interactive methods, which do not require a response from the user’s browser:
- static and dynamic traffic analysis;
- analysis of server response codes;
- monitoring for spikes in requests to specific URLs;
- geofiltering;
- behavioral analysis;
- JA3/JA4 fingerprinting — methods that generate client fingerprints based on TLS handshake characteristics. They can help identify a browser, HTTP library, automation tool, or bot by examining how it initiates a secure connection;
- machine-learning analysis of suspicious requests.
- Interactive methods, which require the browser to complete a check:
- CAPTCHA: a test completed by the user, such as selecting images, entering characters, or moving a slider. Some CAPTCHA systems work invisibly, assessing visitor behavior in the background and presenting a challenge only when bot activity is suspected;
- JavaScript tests: the browser must execute a short script — for example, calculate a value and return it in a cookie. A genuine browser can do this without the user noticing, while most bots cannot;
- HTTP redirects: a mechanism that automatically sends a user from one URL to another. Basic bots may not follow the redirect.
After a visitor passes one of these interactive checks, a cookie is issued to the client to confirm that the check has been completed and the visitor is legitimate.
What Can Slow Down Attack Mitigation?
- Complex and multi-vector attacks. If a rule needs to be changed manually, a straightforward adjustment can be made in the web interface in a couple of minutes and propagated to the filtering nodes within seconds. Complex attacks require more work: engineers first need to understand what they are dealing with, collect data, identify the attack vectors, and adjust the protection settings. If the initial rule does not solve the problem, further analysis and configuration changes are needed, extending mitigation time. In a multi-vector attack, each vector must be addressed separately. If the attacker switches vectors during the incident, the rules may need to be adjusted several times.
- Low-volume application-layer (L7) attacks, such as those generating around 100 RPS. These attacks can blend in with legitimate traffic and require detailed analysis. In such cases, the detection time is usually longer.
We once handled a bot attack targeting an SMS form. A large company was receiving a high volume of requests to send SMS messages through its application. The requests came from malicious bots: the senders entered random, nonexistent phone numbers in the Phone field. The request rate was only around 100 RPS, so the spike was barely noticeable on the overall traffic graph. The attack was detected after the customer reported a sharp increase in attempted SMS sends.
DDoS Protection for Web
- Activate protection in 10 minutes
- 24/7 technical support
What Does an SLA Cover?
A DDoS protection SLA is a service-level agreement that specifies the guaranteed availability of the protected service, expressed as a percentage, and support response times. It does not set a separate mitigation-time target, for the reasons described above.
StormWall’s SLA defines several key terms:
- DDoS attack: an event that causes the protected resource to become overloaded or unavailable and requires protective measures.
- Unavailability: a state in which users cannot access the service or its key functions.
- Scrubbing quality: the effectiveness of malicious traffic filtering while minimizing its impact on legitimate traffic.
Downtime caused by factors outside the provider’s responsibility does not count as an SLA violation. Examples include a failure of the customer’s own services or their internet service provider.
If the SLA is not met, compensation is provided.
The agreement typically specifies support response time explicitly. It does not set separate targets for TTD or TTM — the time to detect and mitigate an attack — but both metrics indirectly affect the availability guarantee. If an attack takes too long to mitigate, the protected resource may become unavailable, and that downtime counts toward the SLA. Customers can monitor attack detection and mitigation in real time through the Client Portal and review the details in reports.
SLA commitments depend on the service plan. For example, StormWall for Web guarantees at least 98% availability and a 3-hour support response time on the Personal plan, compared with at least 99.5% availability and a 15-minute response time on the Enterprise plan.
How to Evaluate Your Response Time Using Client Portal Data
Other companies’ statistics tell you little about how your own service will perform. Every customer has a different traffic profile, detection thresholds, and mix of attack types. The most reliable way to assess response time is to calculate it using data from your own incidents. You can find the information you need in the Client Portal.
- Enable attack notifications. You’ll receive alerts when each incident starts and ends. The interval between these events shows the attack duration.
- Review the attack report. The Reason field shows which rule was triggered. Compare its timestamp with the point where traffic first starts rising on the graph. If there is a significant gap, you may need to lower the threshold.
- Build a sample using Attack History. Filter incidents by time, sort the results, and export them as PDF or CSV reports. The Report Scheduler can also send reports to your email on a set schedule.
- Export data through the API for further analysis.
Keep in mind that traffic graphs use adaptive sampling. When traffic exceeds the 95th percentile, the graph displays the maximum value for the period; under normal traffic conditions, it displays the average. This makes graphs useful for assessing attack peaks. For precise timestamps and values, however, use the table or export the data through the API for analysis in your own systems.
Three Scenarios: What Happens After an Attack Begins?
The exact figures depend on the incident. The following scenarios show which metric is most likely to be affected in each case.
Volumetric attack (L3–L4). The packet rate rises sharply, quickly triggering detection thresholds. Engineer involvement is usually unnecessary. In always-on mode, there is virtually no delay before filtering begins. In on-demand mode, mitigation also depends on how long it takes to redirect traffic for scrubbing — automatically through a DDoS sensor or manually when the customer announces the prefixes.
Low-volume attack (L7). Bots send a relatively small number of requests and imitate legitimate users, so traffic-volume thresholds may not be triggered. Machine-learning-based detection identifies anomalies in IP addresses, URLs, and TLS fingerprints. Most of the time is spent detecting the attack (TTD), which depends largely on the sensitivity of the protection settings.
Multi-vector attack. A threshold is exceeded, and some of the malicious traffic is blocked immediately. However, traffic from other attack vectors may continue to reach the server. An engineer adds filtering rules and checks the results. If the attacker changes the attack parameters, the process is repeated. TTM increases, particularly when the engineer needs additional information from the customer to configure the rules.
When Does an Engineer Step In to Mitigate an Attack?
An engineer gets involved when automated protection cannot handle a complex DDoS attack with the existing configuration. The engineer collects data, analyzes the attack pattern, and adjusts the filtering rules.
There are usually two reasons for an engineer to respond to an incident:
- The monitoring system detects a problem. On-call engineers monitor metrics for incoming requests and the ones that pass through filtering and reach the protected resource. A noticeable increase in the number of requests indicates that the protection settings need to be adjusted and new rules applied.
- The customer contacts support. This often happens during low-volume attacks, when traffic thresholds are not exceeded but malicious traffic is still reaching the protected resource.
What to Do — and What Not to Do — During an Attack
In most cases, you don’t need to do anything. Traffic filtering is automated, and manual intervention will not make it faster. The exception is a protection setup that requires the customer to redirect traffic for scrubbing.
You should contact support if:
- Filtering is active, but the server is still overloaded or returning errors.
- The service is experiencing disruptions, but no attacks appear in the reports.
- You expect a traffic spike — for example, during a sale, a marketing campaign, or a product launch — and want to make sure it is not mistaken for an attack.
- Users are being shown CAPTCHA challenges or cannot access the website.
Do not change server or protection settings in the Client Portal if an engineer is already handling the incident. If you have made any changes, let the support team know in your ticket.
What to Include in a Support Request to Speed Things Up
When submitting a support request, specify its priority, category, product, and subject. Describe the issue and attach a file if relevant. The priority and category help route the request to the right specialist, while a detailed description and screenshot can speed up the investigation.
Include timestamps and specify the time zone. Identify the protected resource and describe what the problem looks like from the user’s perspective. If possible, provide additional technical details. For a website, this could be a section of the web server logs covering the relevant period. For a network or TCP/UDP service, attach a traffic dump — a record of network communication between devices — or monitoring data. Engineers can use this information to create additional filtering rules.
Frequently Asked Questions
How long does it take to mitigate a DDoS attack?
The total time includes attack detection and mitigation. If an engineer needs to get involved, support response time is another factor.
When does the clock start for attack detection and mitigation?
Detection time is measured from the actual start of the attack to the moment a detection rule is triggered (as shown in the Start field of the report). Mitigation time is measured from detection until traffic reaching the server returns to its normal level. For complex and multi-vector attacks, several rounds of rule adjustments may be needed between these two points.
Do I need to contact support when an attack begins?
Usually not. Most attacks are filtered automatically. Contact support if the server is overloaded or users report that the resource is unavailable. Low-volume L7 attacks are harder to detect based on traffic volume alone. StormWall’s machine-learning detection identifies most of these DDoS attacks, but the customer may be the first to notice attacks targeting business logic.
How quickly does support respond?
Response-time guarantees vary by provider, product, and service plan. At StormWall, response times range from 15 minutes to 3 hours.
Where can I find an attack report, and what information does it contain?
Attack reports are available in the Client Portal. A StormWall report includes a graph showing the number of attacks per day, with colors indicating the metric used to detect each attack: PPS, BPS, or RPS.
The Client Portal also provides an attack table listing the protected object, detector, peak value, start and end times, triggered rule, and attack type (L3 or L7). Reports can be exported as PDF and CSV files or sent to your email on a set schedule.
Author: Julya Ilyina, Technical Expert, StormWall
Network Protection from DDoS Attacks
- Activate protection within 10 minutes
- 24/7 technical support