Lessons learned building supervised machine learning into DDoS Protection

Lessons learned building supervised machine learning for DDoS Protection. Data Scientists trained a model to auto-configure security policies and addressed biases in labeled data. Problem statement: automate security policies creation for DDoS protection. Manual intervention doesn't scale due to growing customer numbers and changing network traffic. Human error risk with manual policy creation. Experts visually inspect IP range traffic to create/update security policies. ML model uses time series traffic to predict security policy. Data collection challenges: data freshness and anomalies