Unit 4 wrap-upPut it together

| Topic | Big idea | What the exam asks |
|---|---|---|
| 4.1 The internet | A network of networks; data moves in packets along dynamic routes | Vocabulary, packets, bandwidth time, TCP or UDP, internet vs Web |
| 4.1 Protocols | Open, agreed-upon rules let any device join; that helps it scale | Which protocol does which job; why open matters |
| 4.2 Fault tolerance | Redundant paths let data be rerouted when parts fail | Count paths, minimum failures, single points of failure |
| 4.3 Parallel and distributed | Split work across processors or devices; the sequential part limits speedup | Parallel time, speedup, best split, when more processors stop helping |
Three threads run through everything. Splitting: messages become packets, the internet is many networks, and big jobs become tasks on many processors. Shared rules: open protocols let any device join and let packets be put back together. Redundancy: extra paths make the network fault tolerant, and extra machines let a distributed job survive a failure. And each of these has a cost: resending takes time, redundancy takes hardware, and the sequential portion limits every speedup.
?Wait, which protocol puts packets back in order?
I thought about it. Show me
Traps that cost points
Key relationships for Unit 4
Exam strategy for Unit 4
- Unit 4 is 11 to 15 percent of the multiple choice. No written-response prompt is about it directly, so nearly all its points are in Section I.
- Vocabulary questions: match the word exactly. Protocol, packet, bandwidth, path, routing, redundancy, fault tolerant, speedup.
- Network drawings: redraw them, cross out what failed, and test singles before pairs.
- Timing questions: write the sequential time, the longest task and the sequential part before doing anything else.
- Select two answers: mark each statement true or false on its own first.
- Your Create task can be about a Unit 4 idea, like a program that finds missing packet numbers. The written responses are still about your code, so the networking is the story around it.
- (a) Live stream: UDP, because waiting for lost packets would freeze it. Download: TCP, because the file must arrive complete.
- (b) 3 GB = 3,000 MB = 24,000 megabits. 24,000 ÷ 40 = 600 seconds, which is 10 minutes.
- (c) Sequential: 10 + 30 + 20 + 20 + 10 + 5 = 95 minutes.
- The clips total 80 minutes, so the ideal is 40 per processor: 30 + 10 = 40 and 20 + 20 = 40. Parallel time: 10 + 40 + 5 = 55 minutes.
- Speedup = 95 ÷ 55 ≈ 1.73.
Why?
Brute force in Python confirms 40 is the best possible split of those four clips.
Answer: (a) UDP for the live stream, TCP for the download. (b) 600 seconds, 10 minutes. (c) 55 minutes on two processors, speedup about 1.73.
Stuck? Get a hint
- For (a), ask: is waiting worse, or losing a piece?
- For (b), GB to MB, then × 8, then ÷ 40.
- For (c), split the four clips into two piles of 40.
- A1
- B2
- C3
- D4
Show the answer
B. P has only two connections, P-Q and P-S. Cutting both isolates P, so 2 is enough, and no single failure works: without P-Q, P goes P-S-R; without P-S, P goes P-Q-R. The answer is 2.
- AIts protocols are open, and new networks can join by following them
- BIts protocols are proprietary, and one company approves new networks
- CEvery packet follows one fixed path, and paths never change
- DAll data travels as one large file, and no packets are needed
Show the answer
A. Scalability comes from open protocols that any new network can follow, with no central approval. The other choices describe the opposite of how the internet works: protocols are open, routing is dynamic and data travels in packets.
