Introducing Advanced Rate Limiting

IP addresses no longer serve as a reliable security primitive. Mobile carriers use carrier-grade NAT (CGNAT) to share one address among thousands of subscribers, and distributed bots keep per-IP request volume low precisely to stay under throttling thresholds. Several jurisdictions also treat IP addresses as personal data, which makes replacing IP-dependent security controls attractive on privacy grounds. Rate limiting is one of the products caught up in this shift.

A conventional rate limiting rule combines a filter — usually a path such as /login — with the maximum number of requests a user may make in a time window. Exceeding the threshold triggers an action, typically a block, against that user for a timeout period. Traditional throttling groups requests by IP on the assumption that one address means one user, an approach customers report as ineffective, particularly for authenticated APIs.

Advanced Rate Limiting counts requests on virtually any HTTP request characteristic, independent of source IP. It defends against brute force, scraping and targeted DDoS attacks, whose consequences include sensitive data leakage, account takeover and exhausted back-end resources. Control matters most for APIs, where a single call can set off expensive computation at the origin.

How the mechanism works

The feature ships as part of the Web Application Firewall (WAF) and integrates with Firewall Rules. Rule filters can draw on any HTTP request characteristic: URI, method, headers, cookies and body fields, plus the bot score dynamic field for Bot Management customers. Two response characteristics are also available as triggering conditions: status code and response headers.

Counting dimensions include IP, country, header, cookie, AS Number (ASN), query parameter value and bot fingerprint (JA3). Fields can be used singly or combined, so requests are bucketed whenever their values match. The threshold can instead be expressed as the maximum complexity the origin can handle rather than a maximum request count.

For Enterprise customers, Advanced Rate Limiting removes the traffic caps that previously limited Rate Limiting to a portion of total traffic, and it runs across the entire network, China included.

Web traffic use cases

Because filters can reference any request field, rules can key on specific headers such as User Agent or throttle bots that share an ASN.

Mitigation and counting expressions can now be defined separately. That lets you count requests to /login, for instance, yet block the same user across the whole site. The split is particularly useful when response fields belong in the counting expression — say, counting only requests returning a given response code while blocking a broader set of traffic.

With Advanced Rate Limiting you can count on a path and block across your whole website.

Dynamic fields connect rate limiting to vulnerability detection. A rule can block after a run of consecutive requests flagged as SQLi, activate throttling only for traffic likely to originate from bots (via bot score in the filter), or fire after repeated login attempts with stolen credentials. JA3 fingerprint works as a counting dimension, so the Bot Machine Learning algorithm buckets traffic from bots sharing a fingerprint.

With Advanced Rate Limiting you can use dynamic fields in the rule filter.

API protection

Authenticated API traffic is usually tied to a session carried in a cookie, a header such as x-api-key, or a query value. Advanced Rate Limiting lets you specify where that ID lives and count requests per session regardless of IP — effective against distributed bot attacks that scrape data such as product prices or airline passenger records.

You can group requests based on cookie, header, query and other characteristics.

Rule filters expose both raw and JSON-parsed bodies. A body field with a specific value can be matched using lookup_json_string, which suits GraphQL APIs where one endpoint serves different operations or mutations selected in the body.

You can trigger rules based on request body content.

Complexity-based rate limiting, in beta and available now via API, addresses the fact that request counts misstate serving cost. Some API calls, GraphQL being the prime example, vary widely in processing demand. The origin can estimate each request's complexity, return that estimate with the response, and rate limiting can increment the counter by it; once a complexity threshold in the rule is exceeded, subsequent requests trigger an action such as a block.

Availability

Advanced Rate Limiting is generally available to Enterprise customers on the new Advanced plan. Pro and Business customers cannot use it, though some advantages for those plans are planned. Details of each plan's inclusions are below; the account team or Customer Success Manager can provide more.

Enterprise Core

Enterprise Advanced

Available request fields in filter

Selected standard fields:
URL
Method
Headers
Source IP

All standard fields
Body fields
Account takeover fields
Dynamic fields (including Bot Score*)

Available response fields in counting filter

Response code
Response Headers

Response code
Response Headers

Counting characteristics

IP

IP
IP with NAT awareness
ASN
Country
Headers
Cookie
Query
JA3*

Complexity

No

Yes

Maximum sampling period

10 minutes

1 hour

*requires Bot Management plan

Over the coming months, customer feedback will shape further features. One area already under exploration is automatic traffic profiling that recommends thresholds for rules.