We are currently revamping our website to improve user experience. During this time, some pages or links may not function as expected. We regret any inconvenience and appreciate your patience and continued support.

DDoS Attack Mitigation: Steps to Keep a Site Online

For IT heads keeping public sites online in India or the GCC, DDoS attack mitigation is eight steps in order. They run from a traffic baseline and alerts to filtering in front, WAF rules and rate limits, an escalation path, a rehearsed runbook and a write-up after each attack. Each step names what it stops, who owns it and what it produces, on AWS and on Azure.

DDoS attack mitigation is the work of keeping a site reachable while a flood of junk traffic hits it. The eight steps below put that work in order. They run from a traffic baseline to a write-up after each attack, on AWS or Azure.


Last reviewed 2026-11-16 by Manish Sharma.

This guide is for IT heads and infrastructure leads in India or the GCC whose public websites, portals or APIs earn sales or leads. Begin with those sites. People ask how to prevent DDoS attack trouble, and the plain answer is preparation. No site can stop anyone from sending a flood, but a prepared site absorbs and filters it. The encyclopedia covers the attack itself and how floods differ by layer. Also, how to tell a real flood from ordinary peaks covers spotting one. However, DDoS mitigation here means the work that comes after the spotting.

DDoS Attack Mitigation at a Glance: Eight Steps and Their Owners

Each row also names what the step limits, who owns it and what it leaves behind.

Step What it stops or limits Owner Output
1. Record normal traffic, set alerts Late notice Network or platform lead Baseline, alert list
2. Filtering and a content delivery network in front Floods reaching servers Network lead Each public site with its front door
3. WAF rules and rate limits Request floods on login, search, checkout Application owner with security Rule set per site
4. Hide origin addresses, plan spare capacity Attacks that skip the front door Infrastructure lead Limited origin access, scale settings
5. Escalation path and runbook Lost time finding people IT head Contact sheet, runbook
6. Rehearse with a simulated flood A runbook nobody has used IT head with the provider Rehearsal notes
7. Run the runbook when a flood starts Confusion during the attack On-call lead Incident log
8. Write up each attack, adjust The same gap twice IT head Write-up, changed rules
1
Baseline and filtering

Steps 1 and 2 set alerts and put filtering in front.

2
Rules and capacity

Steps 3 and 4 add WAF rules, rate limits, a hidden origin and spare capacity.

3
Escalation and rehearsal

Steps 5 and 6 agree the runbook and rehearse it.

4
Response and review

Steps 7 and 8 run the runbook, then write up each attack.

Eight steps in order, from a traffic baseline to a write-up after each attack, on AWS or Azure
The work in order, from baseline to write-up.

Steps 1 to 6 happen before any attack.

DDoS Attack Mitigation Before a Flood: Steps 1 to 6

Preparation is how to prevent DDoS attack on website damage, and its order also matters. Signisys’s working view is that filtering in front (Step 2) comes before WAF tuning (Step 3). The basis is that network floods hit below the application, where WAF rules do not reach. The AWS DDoS resiliency whitepaper (no page date shown) lays out by layer. Microsoft Learn on Azure DDoS Protection best practices (page dated 5 March 2026) also says so in advising a WAF alongside DDoS Protection. We last checked these sources on 16 November 2026.

Before a flood in DDoS attack mitigation: a named alert reader, every public site behind a front door, new rules in Count or Log mode, a contact sheet and rehearsal notes
Five things to confirm before any attack.

Step 1: DDoS Attack Detection From a Traffic Baseline

Record normal requests per minute, bandwidth and error rates for each site, split by hour of the week. A flood then stands out against a known shape instead of a guess.

On Azure, set metric alerts on the key DDoS Protection metrics. Use metrics such as whether a resource is under attack and how many inbound packets the service drops. Azure DDoS Protection also profiles an application’s traffic over time. The Azure DDoS Protection overview (page dated 8 July 2025) describes this.

AWS Shield gives attack visibility through the console, the API, and metrics with alarms, as the AWS whitepaper notes. Check which of these your setup includes, then route the alarms to the same place as the Azure ones.

Also name the person who reads an alert at night. An alert with no named reader is a log entry instead of a defense.

Step 2: Put Filtering and a Content Delivery Network in Front

Every public site, portal and API sits behind the cloud’s network filtering and a content delivery network, so the network cuts a flood off before it reaches the origin servers.

On AWS, AWS Shield with Amazon CloudFront at the edge isolates attacks at the source before they reach the origin. On Azure, Azure DDoS Protection covers public addresses. Azure Front Door also adds global load balancing and caching of static content.

However, sites on the firm’s own servers can use these edge services too. The AWS whitepaper describes Amazon CloudFront with AWS WAF in front of origins hosted on-premises. On Azure, Microsoft Learn advises placing well-known public entities in Azure. This also reduces the impact of an attack on on-premises resources.

List every public site with the front door it sits behind. A site missing from that list is a site missing from the defense.

Key Takeaway Put filtering in front of every public site before you tune WAF rules.

Step 3: Set WAF Rules and Rate Limits for Login, Search and Checkout

A Layer 7 flood sends ordinary-looking page or API requests, so request rules filter it, not network filtering.

On AWS, AWS WAF sits on Amazon CloudFront. Its rate-based rules count requests and limit a source that sends too many. On Azure, Azure Web Application Firewall runs on Azure Front Door, and its custom rules can set rate limits. To see where the two layers meet, read what a web firewall does that network filtering does not.

Start with the login, search and checkout paths, as those are the pages that earn the money. Before you block anything, run each new rule in a non-blocking mode. Use Count in AWS WAF (AWS WAF actions page, no page date shown) and Log in Azure Web Application Firewall. Watch for real customers being caught. The Azure rate limiting page (dated 10 November 2026) warns that grouped limits also hit legitimate users in the same group. Then move a rule to blocking only after a quiet week of logs.

Count or Log first

A new rule in Count or Log mode shows what it would block. You can then spot real customers before the rule turns any away.

APIs need more than a WAF, and a separate article explains why a WAF alone leaves APIs open.

Step 4: Hide Origin Addresses and Plan Spare Capacity

Only the front door may reach the servers. Use an allow list to restrict the exposed address space and the listening ports that are not needed. Then an attacker who finds an origin address finds a closed door.

Plan spare capacity too. Microsoft Learn advises designing applications to scale horizontally under amplified load. The AWS whitepaper also lists scaling to absorb application layer traffic among its techniques. Where the site allows it, spread it across more than one location.

Write the scale settings next to the allow list, so the on-call person sees both during a flood.

Step 5: Agree the Escalation Path and Write the Runbook

Name the people inside the firm who own planning and who own action, as Microsoft Learn on Azure DDoS response strategy advises.

Add the provider route. Each cloud offers a DDoS response team. Check whether your setup includes it; where it does not, a support request is the route.

The runbook records who calls whom and in what order, who may change rules, who tells customers and partners, and where the incident log lives. Instead, one page that a tired person can follow beats ten pages nobody opens.

Step 6: Rehearse the Runbook With a Simulated Flood

Microsoft advises simulation exercises, with scale testing, as a normal part of service availability and continuity planning. Use approved test traffic only, and follow the provider’s rules for testing. AWS lists simulated DDoS as subject to its DDoS Simulation Testing policy (AWS penetration testing page, no page date shown). So read that policy before any test. Also, Microsoft names approved testing partners for Azure.

During the rehearsal, time how long it takes to reach each person and to change a rule. Then write down every point where the runbook was wrong or silent.

Those rehearsal notes feed Step 8, so keep them with the other records. To score how ready a site is before a flood, use a scored list of what to have in place before a flood.

DDoS Attack Mitigation While a Flood Runs: Step 7

To mitigate a DDoS attack in progress, confirm that it is a flood and not a busy day. Then open the incident log and start the runbook.

Call people in the order the runbook sets. Escalate to the provider team if the setup includes one. Tighten rate limits or block sources by rule. Also scale out where the settings allow, and tell customers what they will see.

Azure DDoS Protection identifies and mitigates attacks without user action, and it can alert when mitigation starts. That does not remove the runbook, because the application rules, the contacts and the customer message remain with the team.

One last choice exists: blackhole routing, which drops all traffic to the target. However, its trade-off is that the site goes offline for everyone, real customers included. Treat it as a choice for a target that is already down. Agree in the runbook, before any flood, who may order it.

Blackhole routing

Blackhole routing drops all traffic to the target, so the site goes offline for real customers too.

DDoS Attack Mitigation, Step 8: Write Up Each Attack and Adjust the Rules

Write the record within days, while memory is fresh, with the provider team where one is available for post-attack analysis.

The write-up notes when the alert fired against when users noticed, which rules worked and which changed, who was called and how fast, and what the rehearsal missed. It also records the date and who agreed each change.

Changes go back into Steps 1 to 6. For example, add a new alert, move a rule, correct a contact or add a line to the runbook. An attack with no change afterward has taught the team nothing.

Key Takeaway Every attack ends with a write-up and a change to an alert, rule, contact or runbook.

DDoS Attack Mitigation on AWS and Azure, Step by Step

The table sets both clouds side by side for the steps that differ by cloud. Steps 6 and 8 are the same on both.

Step AWS Azure
1. Baseline and alerts AWS Shield metrics and alarms, where your setup includes them Metric alerts on under-attack and dropped-packet metrics
2. Filtering in front AWS Shield with Amazon CloudFront at the edge Azure DDoS Protection on public addresses, Azure Front Door
3. Rules and rate limits AWS WAF on Amazon CloudFront, rate-based rules Azure Web Application Firewall on Azure Front Door, custom rate rules
4. Origin and capacity CloudFront in front of origins, scale to absorb load Allow list for exposed addresses, scale out
5. Escalation Provider response team, if your setup includes it Provider response team, if your setup includes it
7. During a flood Escalate to the provider team, change WAF rules Mitigation without user action, alert when it starts

A firm on both clouds keeps one runbook with a column per cloud.

Signisys provides security services from design to on-call support and uses the network and zero trust work around it. It also partners with AWS and Microsoft. Help putting every public site behind the right defenses is a working session on your public sites against the eight steps, including sites on both clouds or on your own servers, with the result left open. No charge, with a fixed scope.

Book a consultation
Talk through your DDoS runbook

DDoS Attack Mitigation: FAQ

Who owns the steps, IT or the application team?

Both, because the owner column of the first table shows it. Network and infrastructure leads own the filtering and capacity steps. The application owner owns the request rules, and the IT head owns the runbook, the rehearsal and the write-up.

Which step should a small team start with?

Steps 1 and 2 for the sites that earn sales or leads, then Step 5. A baseline, filtering in front and a named contact sheet give a small team the strongest early cover before the team builds the rest.

On what schedule should the runbook be rehearsed?

On a set schedule the firm agrees, and again after big site changes. Microsoft, for example, treats simulation as a normal part of continuity planning. Pick a rhythm the team will keep, then put the date in the runbook.

What should the write-up after an attack record?

The alert time against the time users noticed, the rules that worked and changed. Also whom the team called and how fast, and what the rehearsal missed. Add the date and who agreed each change, so the next reader sees why.

Newsletter

Security Insights, Delivered.

Practical, vendor-neutral analysis of threats, tools, and architecture decisions.

Book a Consultation

A practice principal will contact you within 1 business day.

Thank you! A Signisys principal will be in touch within 1 business day.