Networking Concepts
Networking concepts form the theoretical foundation every technician builds on. This chapter covers the OSI and TCP/IP models, common ports and protocols, IP addressing and subnetting, topologies, and cloud and virtualization fundamentals. Mastering these ideas makes every later hands-on task easier to reason about.
The OSI Model and Encapsulation
The OSI reference model divides network communication into seven layers so that complex problems can be broken into smaller, well-defined pieces. From the bottom up the layers are Physical (Layer 1), Data Link (Layer 2), Network (Layer 3), Transport (Layer 4), Session (Layer 5), Presentation (Layer 6), and Application (Layer 7). A common memory aid is "Please Do Not Throw Sausage Pizza Away." Each layer serves the layer above it and depends on the layer below, which is why a fault at one layer produces predictable symptoms at the layers that rely on it. Layer 1 moves raw bits over copper, fiber, or radio and defines connectors, voltages, and pinouts. Layer 2 packages bits into frames, uses MAC addresses for local delivery, and detects errors with a frame check sequence. Layer 3 adds logical IP addressing and makes routing decisions between networks. Layer 4 provides end-to-end delivery with TCP or UDP, segmentation, and port numbers. The upper layers manage sessions, translate and encrypt data, and present the interface applications use. As data travels down the stack it is encapsulated: each layer wraps the data from above in its own header, so the protocol data unit changes name at each step. The PDU is bits at Layer 1, a frame at Layer 2, a packet at Layer 3, and a segment (TCP) or datagram (UDP) at Layer 4. The receiving host reverses the process, stripping headers as data moves up, a process called de-encapsulation. Mapping devices to layers speeds troubleshooting: hubs and repeaters live at Layer 1, switches and bridges at Layer 2, routers and Layer 3 switches at Layer 3, and next-generation firewalls and load balancers can inspect up through Layer 7. The TCP/IP model condenses these into four layers (Link, Internet, Transport, Application) but maps cleanly onto OSI. Thinking in layers lets you isolate a problem quickly, asking whether the failure is physical, addressing, transport, or application.
Ports and Protocols
Every network application identifies itself with a port number layered on top of a transport protocol, so a single host running many services can keep their traffic separate. Ports range from 0 to 65535 and are split into well-known ports (0 to 1023), registered ports (1024 to 49151), and dynamic or ephemeral ports (49152 to 65535) that clients use as source ports. The Transmission Control Protocol (TCP) is connection oriented: it opens a session with a three-way handshake (SYN, SYN-ACK, ACK), numbers bytes for ordered delivery, acknowledges receipt, and retransmits lost data, which suits web, email, and file transfer. The User Datagram Protocol (UDP) is connectionless and adds almost no overhead, so it fits real-time voice, video, DNS lookups, and streaming where speed matters more than guaranteed delivery. For N10-009 you should know a core set of ports cold. Common TCP ports include FTP 20 and 21, SSH 22, Telnet 23, SMTP 25, HTTP 80, POP3 110, IMAP 143, HTTPS 443, SMB 445, and RDP 3389. DNS uses both TCP and UDP on 53, DHCP uses UDP 67 and 68, TFTP uses UDP 69, SNMP uses UDP 161 and 162, NTP uses UDP 123, and syslog uses UDP 514. Secure variants matter too: LDAPS 636, SMTPS or submission 465 and 587, IMAP over TLS 993, and POP3 over TLS 995. Newer protocols on the exam include SIP 5060 and 5061 for voice signaling and the shift toward encrypted management. Beyond memorizing numbers, associate each protocol with its job: DHCP hands out addressing automatically, DNS resolves human names to IP addresses, NTP synchronizes clocks, and SNMP collects device metrics. Knowing the port, transport, and function lets you write precise firewall rules, read a packet capture, and diagnose why a service is unreachable when a port is blocked or a protocol is misconfigured.
IP Addressing and Subnetting
An IPv4 address is 32 bits written as four dotted-decimal octets, paired with a subnet mask that marks which bits identify the network and which identify the host. CIDR notation shortens the mask to a slash and a count of network bits, so 255.255.255.0 is /24. The number of host bits h gives 2^h total addresses per subnet; subtract two for the network ID and the broadcast address to get usable hosts. Worked example one: for 192.168.10.0/24 there are 8 host bits, so 2^8 = 256 addresses, 254 usable, the network ID is 192.168.10.0, the broadcast is 192.168.10.255, and the usable range is .1 to .254. Worked example two: subnet 192.168.1.0/26. A /26 mask is 255.255.255.192; the interesting octet increments by 256 - 192 = 64, giving blocks 0, 64, 128, 192. The host 192.168.1.100 falls in the .64 block, so its network ID is 192.168.1.64, broadcast is 192.168.1.127, usable range is .65 to .126, and each subnet holds 2^6 - 2 = 62 hosts. Worked example three: you need at least 50 hosts per subnet. Solve 2^h - 2 >= 50, so h = 6 host bits, leaving a /26 mask, which also yields 62 usable hosts with room to spare. VLSM lets you apply different mask lengths within one address space so point-to-point links can use a /30 (2 usable hosts) while user LANs use larger blocks. Private RFC 1918 ranges (10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16) are reserved for internal use and require NAT to reach the internet, while 169.254.0.0/16 is APIPA for link-local self-assignment. IPv6 uses 128-bit addresses written in hexadecimal; FE80::/10 is link-local, FC00::/7 is unique local, 2000::/3 is global unicast, and ::1 is loopback. IPv6 favors SLAAC and eliminates broadcast in favor of multicast.
Topologies, Cabling, and Media
Topology describes how devices are arranged and how data flows, and every design distinguishes physical layout from logical behavior. A star topology wires every node back to a central switch, giving fault isolation because a single failed cable affects only one node; it dominates modern LANs. Mesh topologies connect many nodes directly for redundancy and are common in WANs and wireless backhaul; a full mesh of n nodes needs n(n-1)/2 links. Bus and ring topologies are largely historical, though ring concepts survive in some fiber and metro designs. Point-to-point links join exactly two endpoints, and hybrid designs combine forms, such as a star-of-stars used in hierarchical campus networks. Media selection shapes distance, speed, and interference resistance. Twisted-pair copper is rated by category: Cat 5e supports 1 Gbps, Cat 6 supports 1 Gbps to 100 m and 10 Gbps to about 55 m, Cat 6a supports 10 Gbps to 100 m, and Cat 7 and 8 push higher speeds over shorter distances. All twisted-pair Ethernet runs are limited to 100 meters total, including patch cords. Shielded (STP) cabling resists electromagnetic interference in noisy environments, while unshielded (UTP) is cheaper and common in offices. Fiber optic media carries light rather than electricity, immune to EMI and capable of long distances: single-mode fiber uses a narrow core and laser sources for kilometers-long backbone links, while multimode fiber uses a wider core and LEDs or VCSELs for shorter, cheaper runs inside a building. Connectors to know include RJ45 for copper and LC, SC, and ST for fiber, along with transceivers such as SFP, SFP+, and QSFP that let switches accept different media. Straight-through cables connect unlike devices while crossover cables connect like devices, though auto-MDIX now negotiates this automatically. Matching topology and media to the requirement balances cost, throughput, distance, and resilience, and prevents avoidable physical-layer problems later.
Cloud and Virtualization
Cloud computing delivers computing, storage, and networking as on-demand services over the network, shifting infrastructure from capital purchases to consumption-based operating cost. The three core service models divide responsibility differently. Infrastructure as a Service (IaaS) rents raw virtual machines, storage, and networks, leaving the operating system and applications to the customer. Platform as a Service (PaaS) adds a managed runtime and development stack so teams deploy code without managing servers. Software as a Service (SaaS) delivers finished applications such as email or CRM over a browser, with the provider managing everything beneath. A shared responsibility model defines who secures what at each tier. Deployment models describe where resources live: public cloud uses a provider's shared infrastructure, private cloud dedicates infrastructure to one organization for control and compliance, hybrid cloud connects the two so workloads can burst or fail over, and community cloud is shared among organizations with common needs. Key connectivity and design concepts for N10-009 include the Virtual Private Cloud (VPC), which is an isolated logical network inside a provider; subnets and security groups that segment and filter VPC traffic; and network security lists and gateways that control north-south flow. Organizations link on-premises networks to the cloud with site-to-site VPNs over the internet or with dedicated private circuits such as direct connect services for predictable performance. Virtualization underpins the cloud by abstracting physical hardware. A hypervisor runs multiple virtual machines on one host; Type 1 (bare-metal) hypervisors run directly on hardware for production, while Type 2 hypervisors run atop a host OS for testing. Virtual switches, virtual NICs, and virtual firewalls recreate physical network functions in software. Containers take abstraction further by packaging an application with its dependencies to share one OS kernel, improving density and speed. Elasticity and scalability let capacity grow or shrink automatically with demand, and infrastructure as code provisions all of this from version-controlled templates, aligning cost with actual load and enabling rapid, repeatable deployment.
Last updated: September 2026
La práctica sigue gratis. La guía completa de CompTIA Network+ (N10-009) es el material en sí, explicado de principio a fin — un PDF + EPUB descargable que conservas.