Kona Web Application Firewall Overview - Akamai at RSA Conference 2013

Web Application Firewall (WAF)

RSA® Conference 2013

©2013 AKAMAI | FASTER FORWARDTM

The Cybercrime Landscape in 2013

Attacks have become more sophisticated...

…and easier to carry out

Source: hackmageddon.com/

…industry agnostic...


Moving From Network to Application Layer

Target of Tradi,onal DDoS A2acks

Network Layer

(Layers 3/4)

Application Layer

(Layer 7)

Where increasing number of a2acks are focused


Web Application Firewall Highlights

•  Operates at the network edge – over 100,000 servers •  Inspects requests and responses for malicious content and info leakage •  Inspects packets to protect against attacks such as SQL Injections &

Cross-Site Scripts •  Configurable to log or block activities against policy •  Protects organizations against application layer attacks propagated via

HTTP and HTTPS •  Enables compliance with PCI DSS 1.2 section 6.6 •  Provides advanced rate controls (behavioral based protections) •  Propagates quickly (~30 minutes) •  Configured via portal


Kona Security Solutions 2.0

• ModSecurity Rule Update •  Core Rule Set 2.2.6 •  Legacy CRS support

• Akamai Common Rules •  Based on Akamai’s unique view •  20 – 25% of internet traffic

• Advanced Rate Controls •  Session-ID; Client-IP+User-Agent

• Rule Upgrade Wizard



Appendix & Details


Akamai Intelligent Platform™ Deflecting Network Layer Attacks at the Edge Network Layer attack mitigation §  Built-in protection is “always on” §  Only Port 80 (HTTP) or Port 443 (HTTPS) traffic

allowed on Platform o  All other traffic dropped at the Akamai Edge

•  Attack traffic never makes it onto Platform •  Customer not charged for traffic dropped at Edge

o  Absorbs attack requests without requiring identification o  Requires CNAME onto Akamai Intelligent Platform

Absorbs attacks through massive scale §  ~5.5 Tbps average throughput; up to 8Tbps §  Distribution of HTTP request traffic across 100,000+

servers; 1,100+ networks §  No re-routing, added latency, or point of failure

Examples of attacks types dropped at Akamai Edge §  UDP Fragments §  ICMP Floods §  SYN Floods §  ACK Floods §  RESET Floods §  UDP Floods


Custom Rules Web Application Firewall Description § WAF Custom Rules implemented

in Akamai metadata written by Akamai Professional Services

§ Rules are created and managed in customer portal

§  Rules are then associated with firewall policies and deployed with WAF in 45 minutes

The Result § New rule logic can be built to handle

specific use cases for the customer § Rules can be built that execute when

one or more baseline rules or rate control rules match

§ Output of application vulnerability products can be implemented as “virtual patches”

§  Advanced piping to user validation actions can be achieved (prioritization)


Custom Rules Web Application Firewall Description § WAF Custom Rules implemented

in Akamai metadata written by Akamai Professional Services

§ Rules are created and managed in customer portal

§  Rules are then associated with firewall policies and deployed with WAF in 45 minutes

The Result § New rule logic can be built to handle

specific use cases for the customer § Rules can be built that execute when

one or more baseline rules or rate control rules match

§ Output of application vulnerability products can be implemented as “virtual patches”

§  Advanced piping to user validation actions can be achieved (prioritization)


Adaptive Rate Controls Malicious Behavior Detection

§  Specify number of requests per second against a given URL o Controls requests based on behavior

pattern – not request structure •  Use client IP address, session ID, cookies, etc.

§ Configure rate categories to control request rates against digital properties • Mitigate rate-based DDoS attacks

§  Statistics collected for 3 request phases o Client Request – Client to Akamai Server o  Forward Request – Akamai Server to Origin o  Forward Response – Origin to Akamai Server

§  Statistics collected allow us to ignore large proxies and pick out a malicious user hiding behind a proxy

§  Statistics collected allow for detection of pathological behavior by a client o Request rate is excessive for any stage o Requests causing too many Origin errors


Adaptive Rate Controls Malicious Behavior Detection

§  Specify number of requests per second against a given URL o Controls requests based on behavior

pattern – not request structure •  Use client IP address, session ID, cookies, etc.

§ Configure rate categories to control request rates against digital properties • Mitigate rate-based DDoS attacks

§  Statistics collected for 3 request phases o Client Request – Client to Akamai Server o  Forward Request – Akamai Server to Origin o  Forward Response – Origin to Akamai Server

§  Statistics collected allow us to ignore large proxies and pick out a malicious user hiding behind a proxy

§  Statistics collected allow for detection of pathological behavior by a client o Request rate is excessive for any stage o Requests causing too many Origin errors


Security Monitor (1 of 3)

Timeline of Requests by Hour

Visual Display of Requests by Geography

Requests by WAF Message Requests

by WAF Tag

Requests by WAF Rule ID



Multiple ways to display

request statistics



Requests by Client IP address

Requests by City

ARLs being attacked


Technology

Kona Web Application Firewall Overview - Akamai at RSA Conference 2013