TLS Configuration

TLS does heavy lifting if you let it — the protocol itself is solid, the certificate authorities are mostly trustworthy, the browser ecosystem enforces strict baselines. The defaults on Cloudflare, Vercel, Netlify, AWS CloudFront, and the major managed hosts are also solid; they keep up with cipher hygiene so you don't have to. The bugs cluster on self-managed origins and legacy infrastructure that hasn't been touched since the last major TLS event. Sites still serving TLS 1.0 in 2026, certificates expiring without renewal, HTTPS available but unenforced, weak Diffie-Hellman parameters, missing HSTS — each one is a hostile-WiFi-network away from session hijack. The fix is operational discipline, not new technology.

Several things can go wrong at the transport layer. Missing or short-lived HSTS lets the first request happen over plain HTTP, where a network attacker (coffee shop, captive portal, hotel network) sees the session cookie. TLS 1.0/1.1 are deprecated due to known cryptographic weaknesses (BEAST, POODLE, weak MAC); modern browsers and PCI-DSS reject them, but origin servers still negotiate them when offered. Expired or near-expired certificates produce browser warnings users habitually click through. Weak cipher suites (3DES, RC4, anything with NULL or EXPORT) give the attacker faster offline cracking. Bad certificate chains (missing intermediates) cause some browsers to fail validation. Each is a small misconfiguration; together they decide whether your TLS is real or theatrical.

Session hijack on hostile networks is the headline impact — coffee shop WiFi, hotel networks, conference networks, captive portals all routinely run downgrade attacks against unencrypted or weakly-encrypted traffic. Phishing leverage when users click through cert warnings habitually. Compliance failure: PCI-DSS requires TLS 1.2+ for payment data, GDPR's Article 32 enhanced security requirements imply current cipher hygiene, SOC 2 audits flag deprecated TLS as a control weakness. For B2B SaaS, an enterprise customer's procurement scan that finds TLS 1.0 enabled is the kind of thing that adds three weeks to a deal.

Enable HSTS with `max-age=31536000; includeSubDomains; preload` once you're confident your subdomains can support HTTPS. Submit to the HSTS preload list at hstspreload.org so even first-visit connections are HTTPS. Disable TLS 1.0 and 1.1 at your origin or CDN — every modern toolset has the option. Use modern certificates (Let's Encrypt with autorenewal via Certbot, ACME via your CDN, or your provider's managed certificates). Keep cert expiry monitoring on a calendar — most major outages from 'expired cert' are the absence of monitoring, not the absence of renewal capability. Use a CDN with strong TLS defaults if you can't keep up with hygiene yourself; Cloudflare, AWS CloudFront, and Vercel handle the cipher-suite curation for you. Run securityheaders.com and ssllabs.com periodically — both surface regressions before users notice.