Put this into action
Turn this guide into better conversations with Articuler
Use this guide as the research layer, then turn the next step into a live networking workflow: search by intent, prep for the conversation, and send outreach that is built for replies.
Try the Articuler workflowNetworking rounds reward people who can explain what actually happens when you type a URL and press Enter — layer by layer, packet by packet. Whether you are prepping for a backend, SRE, sysadmin, or network engineering loop, interviewers test the same core ideas: the models that describe how data moves, the protocols that carry it, how names resolve to addresses, and how you debug it all when something breaks.
This guide collects the computer networks interview questions that come up most, grouped by topic, each with a short answer you can adapt into your own words. The highest-value preparation is being able to *talk through* a request end to end and reason about trade-offs, not recite definitions. Where a question has a number — a port, a layer count, a header field — know it cold, because that detail is exactly what separates a confident answer from a vague one.
Use these as conversation starters. The categories below map to how real loops are structured:
- Models: OSI and TCP/IP, and how they map to each other
- Transport: TCP vs UDP, the three-way handshake, ports
- Naming and addressing: DNS, IP, subnetting
- Application layer: HTTP, HTTPS, status codes
- Troubleshooting: the questions that test whether you have actually debugged a network
Network models: OSI and TCP/IP
Almost every networking interview opens here. The model questions are a warm-up, but they also reveal whether you understand *why* layering exists: each layer solves one problem and hands a clean interface to the next, so a change in how bits travel on the wire does not force a rewrite of your web app.
Q: What are the seven layers of the OSI model? From bottom to top: Physical, Data Link, Network, Transport, Session, Presentation, Application. A common mnemonic is "Please Do Not Throw Sausage Pizza Away." Be ready to give one job per layer — Physical moves bits, Data Link handles framing and MAC addressing, Network handles logical addressing and routing (IP), Transport handles end-to-end delivery (TCP/UDP), and the top three deal with sessions, encoding, and the application itself. The OSI model is a conceptual reference, not something any single product implements literally.
Q: How does the TCP/IP model differ from OSI? The TCP/IP model (also called the internet protocol suite) is what the internet actually runs on. It collapses OSI's seven layers into four: Link, Internet, Transport, and Application. OSI's top three layers (Session, Presentation, Application) fold into TCP/IP's single Application layer, and OSI's bottom two fold into the Link layer. OSI is the teaching model; TCP/IP is the engineering one.
Q: Which layer does a router operate at versus a switch? A router works at Layer 3 (Network) — it forwards packets between different networks using IP addresses. A traditional switch works at Layer 2 (Data Link), forwarding frames within a single network using MAC addresses. The distinction matters: routing decisions cross network boundaries; switching decisions stay local.
| Aspect | OSI model | TCP/IP model |
|---|---|---|
| Number of layers | 7 | 4 |
| Origin | ISO reference standard | Practical, evolved from ARPANET |
| Application handling | Split into 3 layers | Single Application layer |
| Primary use today | Teaching and troubleshooting vocabulary | The protocols the internet runs on |
Transport layer: TCP vs UDP
This is the most asked pair in the entire field, and the comparison table below is one you should be able to reproduce from memory. Both live at Layer 4, but they make opposite trade-offs.
Q: What is the difference between TCP and UDP? TCP is connection-oriented and reliable: it sets up a connection, guarantees ordered delivery, retransmits lost packets, and controls flow and congestion. UDP is connectionless and fast: it just sends datagrams with no handshake, no ordering, and no retransmission. As Cloudflare's explainer on UDP puts it, UDP skips the handshake so one computer can simply start sending data — which is why it wins for time-sensitive traffic where a dropped packet beats a delayed one. The formal definitions live in RFC 9293 for TCP and RFC 768 for UDP.
Q: When would you choose UDP over TCP? Real-time, loss-tolerant traffic: live video and voice (VoIP), online gaming, and DNS queries. If retransmitting a late packet would arrive too stale to matter, UDP is the right call. Choose TCP whenever correctness matters more than latency — file transfers, web pages, email, anything where a missing byte corrupts the result.
Q: Explain the TCP three-way handshake. Three steps establish a connection: the client sends SYN, the server replies SYN-ACK, and the client confirms with ACK. After that exchange, both sides have agreed on initial sequence numbers and data transfer can begin. MDN's glossary entry on the TCP handshake is a clean reference if you want to phrase this precisely. A good follow-up answer mentions that this round trip is exactly why connection setup adds latency, and why protocols like QUIC try to reduce it.
| Property | TCP | UDP |
|---|---|---|
| Connection | Connection-oriented (handshake) | Connectionless |
| Reliability | Guaranteed, ordered delivery | Best-effort, no guarantees |
| Speed / overhead | Slower, more header overhead | Faster, minimal overhead |
| Retransmission | Yes, lost packets resent | No |
| Typical uses | Web, email, file transfer | Video/voice, gaming, DNS |
Naming and addressing: DNS, IP, and subnetting
This block tests whether you understand how a human-typed name becomes a route across the internet. Expect questions that chain together — DNS resolution, then IP addressing, then subnetting math.
Q: What is DNS and how does a query resolve? DNS is the Domain Name System, the internet's phone book — it translates a name like articuler.ai into an IP address. A resolution walks a hierarchy: the resolver checks its cache, then queries a root server, then the top-level-domain server (.ai), then the authoritative server for the domain, which returns the record. The behavior is specified in RFC 1035. Mention caching and TTL — it shows you understand why DNS changes are not instant.
Q: What is the difference between an A record and a CNAME? An A record maps a hostname directly to an IPv4 address (AAAA for IPv6). A CNAME is an alias that points one name to another name, which then has to resolve to an A record. A useful detail: you cannot put a CNAME at the zone apex (the bare domain) in standard DNS.
Q: Walk me through subnetting a /24 into four subnets. A /24 gives you 256 addresses (a 24-bit network prefix, 8 host bits). To split it into four equal subnets, borrow 2 host bits, making each subnet a /26 with 64 addresses — 62 usable, since each subnet reserves one network address and one broadcast address. Subnetting is just dividing one IP network into smaller logical pieces; the math is borrowing host bits to create more network bits. Practice converting between CIDR notation and usable-host counts until it is automatic.
Q: What is the difference between a public and a private IP address? Private ranges (10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16) are reserved for internal networks and are not routable on the public internet. Public IPs are globally unique and reachable. NAT (Network Address Translation) is what lets many devices behind one private network share a single public IP.
Application layer: HTTP and HTTPS
If you are interviewing for anything web-adjacent, expect the request-response cycle and the difference HTTPS makes.
Q: What happens when you type a URL and press Enter? The canonical end-to-end question. A strong answer chains the layers: the browser resolves the domain via DNS, opens a TCP connection (three-way handshake), negotiates TLS if it is HTTPS, sends an HTTP request, the server returns a response, and the browser renders it. MDN's HTTP overview is the reference for the application-layer half. Naming each step in order is what scores points here.
Q: What is the difference between HTTP and HTTPS? HTTPS is HTTP layered over TLS, which encrypts the traffic between client and server. It protects against eavesdropping and tampering and authenticates the server via a certificate. The HTTP semantics — methods, headers, status codes — are identical; the difference is the encrypted transport underneath.
Q: Explain a few HTTP status code classes. Group them by leading digit: 2xx success (200 OK), 3xx redirection (301 moved permanently, 304 not modified), 4xx client error (404 not found, 403 forbidden), 5xx server error (500 internal, 502 bad gateway, 503 unavailable). Knowing why a 502 differs from a 503 — bad upstream response versus server overloaded — signals real operational experience.
Troubleshooting questions
These separate candidates who have read about networks from those who have fixed them. Interviewers want to hear a methodical layer-by-layer approach, not a guess.
Q: A user reports "the site is down." How do you diagnose it? Work the stack bottom-up or outside-in. Confirm scope (one user or everyone), then ping to test basic reachability, nslookup or dig to check DNS resolution, traceroute to find where packets stop, curl -v to inspect the actual HTTP exchange, and check whether the issue is the network, DNS, or the application returning a 5xx. Narrating which tool tests which layer is the whole point.
Q: What does `ping` actually test, and when is it misleading? ping uses ICMP to test Layer 3 reachability and round-trip latency. It is misleading when a host blocks ICMP (so a failed ping does not mean the host is down) or when the network works but the application port is closed — ping says "reachable" while the service is still unavailable. Reach for telnet host port or nc to test the actual port.
Q: How would you investigate intermittent latency? Look for patterns: time of day (congestion), specific routes (traceroute/mtr to find a slow hop), packet loss versus pure delay, and DNS resolution time versus connection time (curl -w breaks this down). Distinguishing where the time goes — resolution, handshake, server processing, or transfer — is what a senior answer does.
Next step
Use Articuler to act on what you just read
Start with one concrete goal: investor intros, sales prospects, event meetings, hiring-manager outreach, or expert conversations. Articuler turns that goal into people, prep, and messages.
Start networking with intentFAQ
What computer networks topics should I study first for an interview?
Prioritize the OSI and TCP/IP models, TCP vs UDP, the three-way handshake, DNS resolution, subnetting math, and the HTTP request-response cycle. These five areas account for the majority of questions across SRE, sysadmin, backend, and network engineering loops. Add hands-on comfort with ping, dig, traceroute, and curl for the troubleshooting round.
How technical do computer networks interview answers need to be?
Match the depth to the role and the question. For a definitional question, a crisp two-sentence answer with the load-bearing number (seven layers, three handshake steps, port 443) is enough. For a "walk me through" or troubleshooting question, narrate the steps in order and explain trade-offs — that is where depth wins.
Do I need to memorize port numbers?
Know the common ones: 80 (HTTP), 443 (HTTPS), 53 (DNS), 22 (SSH), 25 (SMTP), 3306 (MySQL). Interviewers rarely quiz you on obscure ports, but blanking on 443 reads as a gap. Memorize the everyday handful and reason about the rest.
What is the most common computer networks interview question?
"What happens when you type a URL into your browser and press Enter?" It is asked constantly because a complete answer touches DNS, TCP, TLS, HTTP, and rendering — letting one question test your grasp of the whole stack. Practice saying it out loud end to end.
Final thoughts
Strong networking answers share a pattern: name the layer, give the specific number or step, then explain the trade-off in plain language. Lock down the five core areas — models, transport, DNS and addressing, HTTP/HTTPS, and troubleshooting — and practice narrating a request from URL to rendered page. For the broader loop, pair this with a technical interview questions refresher, brush up architecture-level thinking with system design interview questions, and if you are targeting infrastructure roles, work through DevOps interview questions too. Day-of nerves and delivery matter as much as content, so it is worth reviewing how to ace an interview before you walk in.
The fastest way to walk in prepared is to know exactly who is on the other side of the table. Articuler helps jobseekers find the hiring manager or engineer running the loop, then builds a Playbook on their background and what they care about — so you can tailor your answers and send a note that gets a reply instead of vanishing into an ATS.