CompTIA Network+ Cheat Sheet: Key Facts
Every key takeaway and common trap from the CompTIA Network+ study guide on one page, grouped by chapter. Tap a lesson link to read the full explanation.
49 facts · 6 chaptersFacts checked 2026-11-16
Chapter 1
Models, Media & Topologies
Key takeaways
- Data is called a segment at the Transport layer, a packet at the Network layer and a frame at the Data Link layer. → The OSI and TCP/IP models: layers, data units and devices
- 10GBASE-T runs about 55 m on Category 6 but the full 100 m channel on Category 6a. → Copper and fiber: categories, distances and connectors
- Single-mode fiber reaches much farther than multimode because its tiny core carries one mode and avoids modal dispersion. → Copper and fiber: categories, distances and connectors
- In 2.4 GHz, only channels five numbers apart, such as 1, 6 and 11, avoid overlapping. → Wireless basics: channels, bonding, antennas and identifiers
- Each switch port is its own collision domain, while all ports in one VLAN share one broadcast domain. → Topologies and network types: domains, mesh, spine and leaf, SAN
Common trap
- The TCP/IP model does not drop OSI layers 5 to 7; it folds all three into its single Application layer. Answers that put them in the Transport layer, or say TCP/IP has no equivalent, are wrong. → The OSI and TCP/IP models: layers, data units and devices
- Do not pick plenum cable for speed, noise shielding or extra distance. It changes only how the jacket burns. Likewise, an injector is called midspan because of where it sits, not because of how many watts it supplies. → Installing cable: plenum ratings and Power over Ethernet
Quick reference
| OSI layer | Data unit | TCP/IP layer |
|---|---|---|
| 5-7 Session, Presentation, Application | Data | Application |
| 4 Transport | Segment (TCP) or datagram (UDP) | Transport |
| 3 Network | Packet | Internet |
| 2 Data Link | Frame | Link |
| 1 Physical | Bits | Link |
| Standard | Approximate power at the PSE |
|---|---|
| 802.3af (PoE) | 15.4 W |
| 802.3at (PoE+) | 30 W |
| 802.3bt, all four pairs | Up to about 90 W |
Chapter 2
IP Addressing, Subnetting & Routing
Key takeaways
- Block size is 256 minus the interesting mask octet; the network address is the block boundary at or below the host and the broadcast is one less than the next boundary. → IPv4 subnetting with the block-size method
- Usable hosts equal 2 to the power of the host bits minus 2, so a /26 gives 62, not 64. → IPv4 subnetting with the block-size method
- A router picks the longest matching prefix first; administrative distance and metric only matter after that. → How routers choose a path: prefix, distance, metric and summaries
- An IPv6 address may use the double colon only once, and fe80::/10 addresses are link-local and never routed. → IPv6: shortening, address types, SLAAC and NAT64
- PAT lets many inside hosts share one public IPv4 address by rewriting source ports, while port forwarding publishes one inside service. → VLANs, inter-VLAN routing and NAT
Common trap
- Distractors often give the first usable host (.65) instead of the network address (.64), or forget the minus 2 and claim a /26 holds 64 hosts. Check which of the four values the question actually asks for. → IPv4 subnetting with the block-size method
- Do not read 169.254.x.x as a DNS or duplicate-address problem. A wrong DNS server never changes the address a host receives, and a duplicate address normally announces itself with a conflict warning; no DHCP reply is the classic cause. Also watch for one-octet changes such as 172.32 or 192.169, which are public. → Private, loopback and APIPA ranges, and traffic types
- Administrative distance is not carried in packets and does not drop by one per hop; that is the TTL field. It is a local preference between route sources, checked only after prefix length. → How routers choose a path: prefix, distance, metric and summaries
- The classic EUI-64 mistakes are forgetting to flip the seventh bit (leaving 00 instead of 02) and inserting FEFF instead of FFFE. Check both before choosing. → IPv6: shortening, address types, SLAAC and NAT64
Quick reference
| Range | Meaning |
|---|---|
| 10.0.0.0/8 | Private (RFC 1918) |
| 172.16.0.0/12 (172.16 to 172.31) | Private (RFC 1918) |
| 192.168.0.0/16 | Private (RFC 1918) |
| 127.0.0.0/8 | Loopback, never leaves the host |
| 169.254.0.0/16 | Link-local (APIPA), self-assigned when DHCP fails |
Chapter 3
Ports, Protocols & Services
Key takeaways
- Moving a cleartext service to a new port hides nothing; only an encrypted protocol such as SSH on TCP 22 fixes exposed credentials. → Well-known ports: secure and cleartext pairs
- DNS answers ordinary queries on UDP 53 but needs TCP 53 for zone transfers and large responses. → DNS and DHCP: transports, records, relays and reservations
- A DHCP client sends from UDP 68 to UDP 67, and clients on another subnet need a relay (IP helper) on their gateway. → DNS and DHCP: transports, records, relays and reservations
- SNMP managers poll agents on UDP 161 and receive traps on UDP 162; only SNMPv3 adds per-user authentication and encryption. → Time and monitoring protocols: NTP and SNMP
- ESP encrypts IPsec payloads while AH only authenticates, and GRE is IP protocol 47, not a TCP or UDP port. → Tunnels, IPsec and QoS markings
Common trap
- If a capture shows passwords on TCP 23, moving Telnet to another port or tunneling it inside plain HTTP changes nothing on the wire. The fix is a protocol that encrypts the session. → Well-known ports: secure and cleartext pairs
- UDP is not chosen for voice because it is secure or because TCP has the smaller header. UDP adds no encryption of its own, and its 8-byte header is the smaller one. It wins because it never waits for a retransmission. → TCP and UDP: handshake, flow control and real-time traffic
- UDP 514 is syslog, not SNMP. A monitoring platform waiting for SNMP alerts listens on UDP 162, and NTP never uses TCP. → Time and monitoring protocols: NTP and SNMP
- Watch for swapped roles: answers claiming AH encrypts while ESP only authenticates, GRE is TCP port 47, or DSCP lives in the 802.1Q tag have each pair reversed. A firewall rule for GRE matches the IP protocol field, not a port. → Tunnels, IPsec and QoS markings
Quick reference
| Job | Cleartext or basic | Secure or preferred |
|---|---|---|
| Remote command line | Telnet TCP 23 | SSH TCP 22 |
| File transfer | FTP TCP 21, TFTP UDP 69 | SFTP inside SSH on TCP 22; FTPS is FTP over TLS (AUTH TLS on 21, or implicit on 990) |
| Directory | LDAP TCP 389 (TLS only via StartTLS) | LDAPS TCP 636 |
| Sending mail | SMTP TCP 25, server to server | Submission TCP 587, client with login |
| Reading mail | POP3 TCP 110, downloads to one device | IMAP TCP 143, synced on the server |
| Logging | Syslog UDP 514 | TCP 514 is the old remote shell, not syslog |
| Identifier | Value |
|---|---|
| GRE | IP protocol 47 |
| ESP / AH | IP protocols 50 / 51 |
| IKE | UDP 500, or UDP 4500 with NAT traversal |
Chapter 4
Network Operations
Key takeaways
- A cable map resolves a wall jack to its patch panel and switch port, a logical diagram shows VLANs and subnets, and a rack elevation shows rack units. → Network documentation and the device lifecycle
- A performance baseline is what turns a raw number such as 45% utilization into a judgment of normal or abnormal. → Monitoring with baselines, SNMP and syslog
- Syslog severity 0 is Emergency, the most severe, and 7 is Debug, the least. → Monitoring with baselines, SNMP and syslog
- RPO limits how much data you can lose and drives backup frequency; RTO limits how long you can be down. → Disaster recovery and high availability
- A rollback plan must be written and approved before a change is implemented, not after it fails. → Change management and configuration baselines
Common trap
- DHCP snooping is not IPAM. Snooping blocks rogue DHCP servers and tracks current leases, but it does not plan subnets or reclaim static addresses left behind by retired hosts. → Network documentation and the device lifecycle
- Syslog numbers run backward from intuition: a higher number is less severe. Setting a device to log at level 7 sends every debug message to the collector and can flood storage and the management link. → Monitoring with baselines, SNMP and syslog
- When a question says every frame, including errored frames, with no added switch load, a mirror or RSPAN session is wrong; only a passive TAP fits. Flow data has no frames to examine at all. → Seeing traffic: flow data, TAPs, mirror ports and availability checks
- A 15-minute data-loss limit is an RPO, not an RTO. Backing up more often shrinks possible data loss but does not make a restore any faster. → Disaster recovery and high availability
Quick reference
| Document | Answers |
|---|---|
| Cable map with a labeling scheme | Which patch panel position and switch port a wall jack lands on |
| Logical diagram | VLANs, IP subnets, gateways and traffic paths |
| Physical diagram | Where equipment and media physically sit |
| Rack elevation | Which device sits in which rack unit, for space, power and airflow planning |
| IPAM | Every subnet and address allocation; prevents duplicates and reclaims addresses |
| Term | Meaning |
|---|---|
| RPO | Most data you may lose, measured back from the incident; sets backup frequency |
| RTO | Longest the service may stay down; improved by faster restore or standby gear |
| MTBF | Average operating time between failures (reliability) |
| MTTR | Average time to repair a failed device (maintainability) |
Chapter 5
Network Security
Key takeaways
- A vulnerability is the weakness, a threat is who or what could use it, an exploit is the technique, and risk combines likelihood and impact. → Core security terms: CIA, risk vocabulary and zero trust
- In 802.1X the client is the supplicant, the switch port or AP is the authenticator, and RADIUS is usually the authentication server. → Identity and access: AAA, 802.1X, RADIUS, TACACS+, MFA and RBAC
- ACLs are read top down, stop at the first match and end with an implicit deny. → Read the chapter
- The gateway's IP and another host's IP mapped to the same MAC point to ARP poisoning; a correct name resolving to a wrong address points to DNS poisoning. → Recognizing network attacks from their symptoms
- Two knowledge items, such as a password and a security question, are still one factor and are not MFA. → Identity and access: AAA, 802.1X, RADIUS, TACACS+, MFA and RBAC
Common trap
- A hash protects integrity, not confidentiality: anyone can still read the file. And accounting is never a CIA answer, because it belongs to AAA. → Core security terms: CIA, risk vocabulary and zero trust
- RADIUS and TACACS+ are often swapped in distractors. Remember that TACACS+ is the one that encrypts the full packet and separates authorization from authentication. → Identity and access: AAA, 802.1X, RADIUS, TACACS+, MFA and RBAC
- A rogue AP is any unauthorized access point, such as an employee's home router. It becomes an evil twin only when it impersonates the real network's SSID. → Recognizing network attacks from their symptoms
Quick reference
| Term | Meaning |
|---|---|
| Vulnerability | The weakness itself, such as unpatched firmware |
| Threat | The actor or event that could use the weakness |
| Exploit | The specific technique or code that uses it |
| Risk | Likelihood of exploitation combined with its impact |
| Symptom | Attack | Defense or key note |
|---|---|---|
| Gateway IP and another host's IP map to the same MAC | ARP poisoning | ARP has no authentication; the attacker gains an on-path position |
| Correct hostname, fraudulent site, cached record points elsewhere | DNS poisoning | DNSSEC signing for origin and integrity |
| Switch floods unicast out every port | MAC flooding | Port security limits MACs per port |
| Frame with two 802.1Q tags reaches another VLAN | VLAN hopping by double tagging | Native VLAN set to an unused VLAN |
| Cloned corporate SSID with a stronger signal and a fake login page | Evil twin | Impersonating the real SSID is what defines it |
Chapter 6
Network Troubleshooting
Key takeaways
- If a theory is not confirmed, establish a new theory or escalate; only a confirmed theory moves on to a plan of action. → The troubleshooting methodology, step by step
- Late collisions on one end and runts and CRC errors on the other point to a duplex mismatch. → Reading interface counters and port states
- Output drops with zero CRC, runt and giant errors mean congestion, not a bad cable. → Reading interface counters and port states
- A split pair passes a basic continuity test, so only a certification tester that measures crosstalk will find it. → Cable faults, distance limits, PoE budgets and the right tester
- Steady delay with large swings in packet arrival times is jitter, and a Wi-Fi analyzer cannot see non-802.11 interference that a spectrum analyzer can. → Wireless and voice quality problems
Common trap
- Questioning the user and reproducing the fault are part of step 1, so the next step is a theory, not a fix. And writing down which VLANs and devices a swap will disrupt is step 4, the plan of action, not verification. → The troubleshooting methodology, step by step
- Re-enabling an err-disabled port without fixing what triggered it only lets the switch disable it again. Treat it as a symptom of a violation, not a typo in the configuration. → Reading interface counters and port states
Quick reference
| What you see | What it means |
|---|---|
| Late collisions and FCS errors on one side, runts and CRC errors on the other; ping fine, big transfers crawl | Duplex mismatch; set both ends to match, ideally both on autonegotiation |
| Rising output drops, zero CRC, runts and giants, utilization near 100% at peak | Egress congestion; add capacity or QoS |
| Rising CRC or FCS errors alone | Corruption on the medium, such as a damaged cable or EMI |
| Port err-disabled | The switch shut it for a violation such as port security, BPDU guard or flapping; remove the cause first |
| Port administratively down | A person shut it down |
| Port suspended | Link-aggregation settings do not match the rest of the bundle |