[reference], with grounding in the Linux networking already covered (network fundamentals, SSH). This is the operational protocol knowledge a DevOps engineer debugs with, not a CS networking course.

The OSI model — a debugging map

The OSI model layers networking so you can reason about where a problem is. You rarely need all seven, but the mental model of “which layer is broken?” is the point:

LayerJobYou debug here when…
7 ApplicationHTTP, DNS, SSH, SMTPthe app protocol misbehaves (a 500, a bad DNS answer)
4 TransportTCP, UDP — ports, reliabilitya port’s closed, a connection won’t establish
3 NetworkIP, routingpackets don’t reach the host, routing/subnet issue
2/1 Link/PhysicalMAC, cables, Wi-Fihardware/L2 (rare in cloud)

The practical takeaway: when something’s broken, isolate the layer. “Can I resolve the DNS name?” (L7) → “can I reach the IP/port?” (ping/telnet/nc, L3/L4) → “is the app responding?” (L7). Most cloud debugging is DNS, ports/firewalls (L4), or the app (L7).

TCP vs UDP (L4): TCP is connection-oriented and reliable (handshake, ordering, retransmission) — HTTP, SSH, databases. UDP is fire-and-forget, no guarantees but low overhead — DNS queries, video, metrics. Ports identify a service on a host (80 HTTP, 443 HTTPS, 22 SSH, 53 DNS).

DNS — the internet’s phone book

DNS resolves names (example.com) to IPs. It’s the source of a huge share of “it’s down” incidents, so the record types matter:

RecordMapsUse
A / AAAAname → IPv4 / IPv6the basic address lookup
CNAMEname → another namealiasing (www → the apex)
MXdomain → mail serveremail routing
TXTarbitrary textverification, SPF/DKIM/DMARC (below)
NSdomain → its nameserversdelegation

Two operational realities: TTL (time-to-live) controls how long a record is cached, so DNS changes aren’t instant — lower the TTL before a planned change. And DNS propagation delays are why a fresh deploy or cert sometimes “isn’t live yet.”

HTTP / HTTPS

HTTP is the request/response protocol of the web: a method (GET/POST/…), a path, headers, and a body; the response carries a status code (2xx success, 3xx redirect, 4xx client error, 5xx server error) — reading status codes is first-line debugging. HTTP/2 and HTTP/3 add multiplexing and (in HTTP/3) run over UDP for lower latency, but the semantics are the same.

HTTPS is HTTP over TLS. The essentials a DevOps engineer must own:

  • TLS provides encryption (eavesdroppers can’t read traffic), integrity (it can’t be tampered with), and authentication (the server is who it claims, via a certificate signed by a trusted CA).
  • The handshake negotiates keys before any HTTP flows (asymmetric crypto to exchange a symmetric session key — see cryptography).
  • Certificates expire — an expired cert is a classic, embarrassing outage. Let’s Encrypt + automated renewal (certbot, or a reverse proxy like Caddy that does it automatically) is the standard fix. Certs and their private keys are secrets.

SSH

Secure Shell — encrypted remote access and the transport for tools like Ansible. Operationally: prefer key-based auth over passwords (a keypair; the public key goes in the server’s authorized_keys), disable root/password login, and use the SSH config and agent for convenience — the hands-on details are in Basic SSH Config.

Email protocols (the roadmap’s mail branch)

Rarely built from scratch, but worth recognizing: SMTP sends mail; IMAP/POP3 retrieve it. The deliverability trio lives in DNS TXT records and exists to stop spoofing:

  • SPF — lists which servers may send mail for your domain.
  • DKIM — cryptographically signs outgoing mail so the recipient can verify it wasn’t forged.
  • DMARC — ties SPF+DKIM together and tells receivers what to do with mail that fails (quarantine/reject), plus reporting.

If you ever send transactional email from an app and it lands in spam, misconfigured SPF/DKIM/DMARC is almost always why.

↑ Back to 08 — Networking & Web Servers