DDoS Mitigation: How to Measure Attacks and What to Ask Before You Choose

6 min readPacketStream

DDoS mitigation means telling attack traffic from legitimate traffic. We cover L3/L4 and L7 attacks, bps, pps and rps, Always-On vs On-Demand, WAFs, questions to ask first and the order of response.

On this page
  1. What is DDoS mitigation?
  2. L3/L4 vs L7 attacks
  3. Reading bps, pps and rps
  4. Always-On vs On-Demand
  5. WAF and DDoS mitigation
  6. Questions to ask first
  7. When an attack starts

Compare DDoS mitigation services and the first things you see are mitigation capacity in Tbps, Always-On and L7 protection. But if you do not know which gives out first when an attack arrives, your link, your equipment or your application, those terms cannot serve as a basis for comparison. We will go through what mitigation does, the units for reading attack size, the questions to ask before you choose and the order of response during an attack.

What is DDoS mitigation?

DDoS mitigation filters out attack traffic while letting legitimate users' requests reach the service as they are. A DDoS (Distributed Denial of Service) attack concentrates traffic from many sources on a single target so that legitimate users cannot use the service. Because the sources are scattered, blocking a handful of IP addresses does not end it.

So the hardest part is telling attack traffic from legitimate traffic. If visitors rushing in for an event are treated as an attack and blocked, the mitigation itself causes the outage. Cloudflare describes cloud-based mitigation as four stages.

  • Detection: distinguishes an attack from a legitimate surge in traffic.
  • Response: drops attack traffic and absorbs the rest.
  • Routing: breaks the remaining traffic into manageable chunks and routes it.
  • Adaptation: learns recurring attack patterns to prepare for the next attack.

How L3/L4 attacks differ from L7 attacks

L3/L4 attacks and L7 attacks try to exhaust different resources. AWS groups DDoS attacks into three categories.

  • L3 volumetric attacks: fill the capacity of the target network or resource. UDP reflection and amplification attacks are typical.
  • L4 protocol attacks: abuse the way a protocol works. A SYN flood exhausts the connection state resources of servers, load balancers and firewalls.
  • L7 application attacks: send large numbers of requests that the service accepts as normal. An HTTP flood is typical.

Even with bandwidth to spare, packet processing capacity or connection state resources can run out first, and an L7 attack is effective with few requests. How L7 attacks are picked out is covered in the L7 DDoS Protection article.

Three units for reading attack size: bps, pps, rps

Attack size is read in three units, bps, pps and rps, and each unit has a different resource that hits its limit first. AWS Shield Advanced also records L3/L4 attacks in bits and packets per second and L7 attacks in requests per second.

UnitWhat it measuresWhat hits its limit first
bpsBits per secondBandwidth of links and ports
ppsPackets per secondPacket processing in routers, firewalls and servers
rpsRequests per secondWeb servers, APIs and databases

Think of a parcel sorting center: bps is the total weight of the freight, pps is the number of boxes and rps is the number of order slips. At the same total weight, the sorter jams first when the freight arrives split into many small boxes. Network equipment also pays a processing cost for every packet, so pps can cross its limit first even when bps is below the capacity of the link.

That is why mitigation capacity in Tbps (bps) alone does not tell you how well a service withstands a flood of small packets or a flood of L7 requests. When comparing mitigation services, ask about pps and the scope of L7 handling as well.

The difference between Always-On and On-Demand

Always-On and On-Demand differ in whether traffic passes through the mitigation network in normal operation.

Always-On

  • Traffic designated for protection passes through the mitigation network in normal operation too
  • No waiting for the route to switch to the mitigation network during an attack
  • Depending on the provider's architecture, latency in normal operation may increase
  • Check L7 support and protocol coverage separately

On-Demand

  • Connects directly in normal operation and changes the route, via BGP for example, when an attack is detected
  • Check how traffic is monitored in normal operation and what triggers the route switch
  • Until the route changes, the attack reaches your servers as it is
  • If monitoring in normal operation is volume-based, a low-volume L7 attack may not meet the switch condition

Cloudflare regards Always-On, which inspects traffic in normal operation too, as the most comprehensive approach. With On-Demand, some providers change the route only after the customer contacts them, so check how the switch works in advance. PacketStream DDoS Mitigation uses Always-On by default, and On-Demand is also available.

How a WAF relates to DDoS mitigation

A WAF and DDoS mitigation partly overlap on HTTP floods, but they cover different ground. A WAF (Web Application Firewall) is an L7 defense that filters HTTP traffic with rules to stop vulnerability attacks such as SQL injection and XSS. Cloudflare also explains that a WAF is not a tool designed to stop every type of attack.

CriterionWAFDDoS mitigation
Traffic it looks atHTTP and HTTPS requestsPackets and connections, and L7 protocols depending on scope
Attacks it mainly stopsVulnerability attacks such as SQL injection and XSSAttacks that exhaust bandwidth, connection state and processing capacity
Hard to cover aloneL3/L4 attacks that fill the link, non-HTTP protocolsVulnerability attacks carried in a single request

A WAF on the inside of the service link is cut off along with everything else once the link is full, and it cannot see the traffic of a game server that uses its own UDP protocol. Conversely, DDoS mitigation alone can miss vulnerability attacks that do not show up as volume, so web services commonly run the two together.

Questions to ask before choosing DDoS mitigation

Before you choose, look at where your own service is weak ahead of mitigation capacity.

  1. What are you protecting? Write down the IP ranges, ports and protocols to protect. The L7 protection you need depends on whether the service uses HTTP or its own protocol, such as a game, VoIP or VPN protocol.
  2. How much traffic is normal? The normal and peak values of bps, pps and rps are the basis for thresholds and for judging false positives.
  3. Which layers does it stop? Check whether it validates your service's protocol and whether the scope of L7 protection differs by plan.
  4. Where is the attack filtered? An attack larger than your link's bandwidth fills the upstream link first, so equipment on the inside of the link cannot stop it.
  5. How does the route change? Look at the deployment mode, whether the switch is automatic or manual, and the change in latency. Mitigation that analyzes sessions may need symmetric routing.
  6. What can you see during an attack? Check alerts, attack reports, block history, the procedure for false positive exceptions and the support channel.

The order of response when an attack starts

When you suspect an attack, move in the order below. Having the order settled reduces the chance of hastily lowering server thresholds and blocking legitimate users as well.

  1. Confirm the symptomsCompare against your normal values to see whether the link, the equipment or the application hit its limit first, and check whether it is a legitimate surge.
  2. Narrow the scopeNarrow down the attack layer and the target IP, port, protocol and URL, and choose rules that block fewer legitimate users.
  3. Apply mitigationNotify your mitigation provider and, with On-Demand, confirm that the route has switched. For L7 attacks, apply a challenge or rate limit centered on the targeted paths.
  4. Review afterwardClean up false positives and temporary rules, and realign your normal values, thresholds and contact procedures.

AWS also advises that a value appearing in a detection metric is not enough to conclude an attack is under way, and that you should look further into the state of your resources. This is why the first step is confirming before blocking.

The terms we saw at the start, mitigation capacity, Always-On and L7 protection, become criteria for comparison only once you know where your service is weak. If your normal values tell you what gives out first, the mitigation you need and the first place to look during an attack are decided together.

View DDoS Mitigation

More insights

All insights

Ready to get started?

Tell us about your workload and we will help you choose the right products and plans.

  • Talk to our team

    Questions about anything you read here? Our team will help.

    Contact Sales
  • Insights

    Where PacketStream is heading, what we learn running the network, and the latest product news and notices.

    All insights

Privacy settings

Manage what you allow

Necessary cookies and similar technologies (such as local storage) are always active. You can choose the rest below and change your choice at any time from Cookie settings at the bottom of the site. See our Privacy Policy.

Necessary

Always active

Required for the site to work and stay secure. They cannot be turned off.

Examples: Theme setting, consent record, contact form security check (Cloudflare Turnstile)

Analytics and performance

Optional

Shows how visitors arrive and use our pages so we can improve the site.

Examples: Google Analytics, Microsoft Clarity

Marketing

Optional

Measures how our ads perform and helps show more relevant ads on other sites.

Examples: Google Ads