HTTPS & TLS in 2016: Security practices from the front lines

No ratings

Presented at AppSecUSA 2016 by

Kenneth White

Eric Mill

Implementing strong security for Internet‐facing services has grown more challenging and more complex over the past two years. With protocol‐level vulnerabilities like FREAK, BEAST, CRIME, POODLE, & LOGJAM, Ops teams are forced to reevaluate long‐held assumptions about foundation system network code. What are the right tradeoffs between modern network security requirements versus widespread legacy client and user interoperability? How do we apply these to traditional Apache and Nginx servers, mobile app web services, and non‐browser infrastructure like libcurl, proxies, API endpoints, and load balancers? And what's the deal with Curve25519, ChaCha/Poly1305, LibSodium, BoringSSL, and LibreSSL? Here, we present a practitioner's crash guide to modern site and web service endpoint encryption using HTTPS. We cover the "TLS 101" (and 201) fundamentals of certificates: ECDSA vs RSA, 2K vs 4K, ephemeral Diffie‐ Hellman (elliptic curve versus static), Domain Validation vs Extended Validation. We'll talk about intermediate and root authorities (and why Superfish is such a problem), and then look at some best practices around https including certificate transparency (CT), pinning (HPKP), and strict transport security (HSTS). Lastly, we'll give updates from the OpenSSL 1.1 audit, and point to well curated configuration guides and recipes for https and TLS.