A network exchanges information through agreements called protocols. To understand a query, distinguish the link, network configuration and application. The link connects an interface by cable or radio; configuration enables destinations to be located; the application exchanges messages for a task. These are related dependencies, not equivalent outcomes. A WiFi connection can exist without Internet access, and a local service can work without leaving the classroom network.

Diagnosis starts with a precise symptom: no link is established, a destination fails, a name cannot be resolved or the application returns an error. Each observation narrows subsequent hypotheses. An assigned address is not proof of a correct address, and a visible page may contain old data. Check the relevant dependency instead of restarting everything without recording changes. The laboratory is isolated and authorised; its tests do not modify the school’s general infrastructure. This boundary allows one condition to be changed, observed and restored to its starting state.

Client and server describe roles in an exchange: the client requests and the server provides a service. A computer may perform both roles in different processes. A switch communicates between devices within a link-layer network, an access point allows wireless association and a router connects networks. A domestic enclosure may combine these functions and additional services without making their tasks identical.

Topology represents the devices, links and functions of the network actually used. Drawing a cloud labelled Internet for everything hides the dependencies of a local query. Follow the pathway from client to server and ask whether it really crosses into another network. An ESP32 station may join an access point as a station or provide its own access point, depending on the prepared firmware. Consult and record the mode rather than guessing from the board’s appearance. A simulated server supports reasoning about services but does not establish a physical sensor connection.

An IP address identifies an interface within its network context. To decide which destinations are considered local, the computer also uses the mask or prefix: the address alone does not express the whole relationship. In the IPv4 scenario 192.168.50.0/24, the mask is 255.255.255.0, and addresses ending in .10 and .21 can share a subnet. In this conventional /24, .0 identifies the network and .255 is the broadcast address; neither is assigned to clients. Different interfaces need unique addresses within that context so identification is unambiguous.

When the destination belongs to another network, the gateway provides a route towards it. DNS serves another function: it connects names with addresses. Querying a known local numeric address therefore does not require name resolution, whereas querying a prepared name depends on that resolution. DHCP can provide configuration automatically, but receiving values does not prove that they match those intended. Each function addresses a different part of the pathway.

The authorised plan makes the observed configuration meaningful: address, mask, gateway and DNS are recorded from the system and compared with that plan. This checks a particular relationship between interfaces and destinations instead of copying numbers without understanding them. A private address acts as neither a password nor a guarantee of anonymity. The scenario supports learning; other networks may use different prefixes or IPv6 and require their own authorised values.

A wired link depends on compatible cable, connectors, interfaces and equipment. WiFi adds radio conditions such as distance, obstacles, interference and occupancy, as well as authentication. Signal bars describe reception approximately, not server availability or guaranteed speed. The advertised link rate also differs from useful transfer throughput: protocols, workload and devices influence the result.

Latency describes delay in an operation and should not be confused with data transferred per second. Compare media using the same service and task, document location, device and active interface, and record what is actually observed. Changing client, distance and server together prevents attributing a difference solely to cable or WiFi. Two active interfaces may follow different routes and confuse the test. Use the prepared configuration and avoid overloading the network. A classroom comparison supports conclusions limited to its conditions, rather than declaring one medium universally better for every use.

An address locates an interface, but an application also needs protocol, port and resource. In http://192.168.50.10:8000/panel, HTTP specifies the protocol, 8000 the destination port and /panel the path. This is a laboratory example, not an address to copy into every network. A process must be listening and access must be permitted. The panel and reading may have different paths and authentication conditions.

HTTP 401 indicates missing valid authentication; 403 expresses forbidden access and 404 means the resource was not found on the responding service. None means absent WiFi. A timeout without a response needs further tests and must not automatically be equated with a particular HTTP status. Ping is another check: it may be filtered and even a reply does not guarantee a web service. Test the authorised application task. A valid reading needs value, unit and update context as well as a received response. Service secrets do not appear in the submission.

Dependency-based diagnosis uses what has already been checked to choose the next test. A failed link, an address outside the plan and a stopped service are different situations even if each produces a page that does not load. Distinguish them by observing interface, configuration and application availability with the environment’s tools. There is no need to repeat an entire ritual when specific evidence points to a relevant starting point.

The teacher prepares one fault at a time and restores configuration before the next test. Record symptom, hypothesis, expectation, action, observation, correction and final verification. A discriminating test separates at least two explanations; changing several parameters hides which mattered. If an authorised restart restores service without identifying the cause, say so. Preserve a revised hypothesis when evidence contradicts it. Do not explore other destinations or inspect classmates’ traffic. Simulation supports diagnostic reasoning but must be declared and does not establish physical configuration.

Network documentation combines topology, configuration plan and incident record. The drawing shows connections; the table preserves roles and interfaces; the record explains how a hypothesis changed. They must describe the same setup and version. A screenshot of the final outcome does not show which check ruled out a cause or allow the procedure to be repeated from the initial state. Access secrets remain outside shared submissions.

A useful guide states prerequisite, authorised action, expected outcome and next decision if it differs. Another pair can apply it to a new case without memorising addresses. Transfer is checked by explaining which dependency the first test examines and which result would change it. Record equipment, simulation and test-condition limitations. If the setup changes, update the drawing before reusing conclusions. Verify improvement in both access and the query task; restoring a link alone does not establish that the service again supplies useful information.

In an IoT query, opening the panel checks only part of the pathway. A useful reading combines value, unit and temporal context. The device may record the measurement time and the server the reception time: each describes a different operation. Age becomes meaningful when we know which record it refers to and how it is calculated. Comparing times from unsynchronised clocks may produce an incorrect interpretation. The physical unit also needs support from the sensor record; the number alone does not determine it.

Visible information may persist even when it stops being renewed. The client may retain the screen and the server the last reading after the sender stops. Panel access, a query response and actual data renewal are therefore distinguished. An advancing timestamp or sequence provides evidence about the record it represents; its limits depend on the prepared mechanism. A sequence shows succession, but does not by itself represent a duration. A repeated value does not establish a fault either, because the measured quantity may remain stable.

Diagnosis connects these observations with a hypothesis and a check. If the panel opens but the record does not advance, browser connectivity alone does not explain where the pathway has stopped. Comparing the available sender, server and client records helps narrow down the problem. Changing one condition and repeating the query reveals whether the intervention had the intended effect. Identifying when we use real hardware, a simulated server or fictional data sets the limits of what was checked and which claims remain unresolved.