Chapter 46 — Engineering capstone II
Lesson map
Lesson 46 published. Map, example, EX136.
flowchart TD
N0["Learning outcomes"]
N1["46.1 Completion"]
N2["46.2 Measures"]
N3["46.3 Criteria comparison"]
N4["46.4 Limitations"]
N5["Interpreting the availability figure"]
N6["Unintended effects"]
N7["Cost against outcome"]
N0 --> N1 --> N2 --> N3 --> N4 --> N5 --> N6 --> N7
N7 --> EX["Worked example"]
EX --> Q["Exercise EX136"]
Logical drawing
flowchart LR
N0["Learning outcomes"]
N1["46.1 Completion"]
N2["46.2 Measures"]
N3["46.3 Criteria comparison"]
N4["46.4 Limitations"]
N5["Interpreting the availability figure"]
N6["Unintended effects"]
N7["Cost against outcome"]
N0 --> N1 --> N2 --> N3 --> N4 --> N5 --> N6 --> N7 --> Q["EX136"]
[Learning outcomes]
|
v
[46.1 Completion]
|
v
[46.2 Measures]
|
v
[46.3 Criteria comparison]
|
v
[46.4 Limitations]
|
v
[Interpreting the availability figure]
|
v
[Unintended effects]
|
v
[Cost against outcome]
|
v
[Worked example] --> [Exercise EX136]
Learning outcomes
You will (1) complete implementation; (2) measure outcomes; (3) compare with criteria.
46.1 Completion
By the end of the capstone, all eight offices are migrated to the six-zone model behind their own firewalls, the CCTV supplier reaches cameras only through the jump hosts, and the three offices with funded second links have redundant tunnels. Kofi checks completion against the original scope rather than declaring it done when work slows: each office has a signed-off migration record, a passing regression run and an entry in the operational routines. Two minor items, a label update and a diagram correction, are listed as open with owners.
46.2 Measures
The measured outcomes are: reachable zone pairs down from thirty to seven; unexpected denies settled at around four per week across all offices; VPN availability of 99.97 percent over three months; added latency within the one-millisecond budget in both lab and live measurements; and a change failure rate of four percent. Each figure is linked to its source data in the dossier, so a reader can check how it was produced. Reporting the measures with their sources matters more than the numbers themselves, which would be unconvincing without them.
46.3 Criteria comparison
Kofi compares each result against the criteria agreed before measurement. Guest isolation is verified in both the lab and production. All eight requirements have passing evidence in the traceability table. Availability of 99.97 percent exceeds the 99.9 percent target. Latency is within budget. All four criteria are met. He presents this as a simple table with the criterion, the target, the result and a link to evidence, avoiding persuasive language, so the comparison speaks for itself and examiners can verify each line independently.
46.4 Limitations
Meeting the criteria does not make the results universal. The availability figure covers only three months, which may not include rare events such as a long provider outage. The project involved one council, so results may differ in organisations with different sizes, suppliers or skills. Some measurements, like the incident reconstruction, were possible only in the lab. Kofi states each limitation next to the result it affects rather than in a separate section that readers might skip, which helps readers judge how far the findings can be trusted and applied elsewhere.
Interpreting the availability figure
VPN availability of 99.97 percent over three months means total outage across all offices of roughly forty minutes per office over the period. Most of that came from a single provider fault at one branch before its second link was installed. Since the three offices with second links had no measurable outage, Kofi reports availability separately for single-link and dual-link sites. Averaging across all offices would hide the difference that the project's investment made. He also notes that one provider fault in a short period is not enough to estimate long-term availability accurately for any single site.
Unintended effects
Evaluation should look for effects nobody planned. Kofi checks service desk categories before and after migration. Tickets about network access rose briefly in the first week at each office, then fell below the old level, partly because clearer zone documentation helped staff and engineers understand where devices belonged. One unintended negative effect appears: a few staff began using personal mobile hotspots to reach blocked services. Kofi recommends a clear request process for new flows and a short communication explaining why certain connections are blocked, since workarounds of this kind undermine the controls the project built.
Cost against outcome
The council spent money on firewalls, second links, staff time and Kofi's placement. He presents costs alongside outcomes: the reduction in reachable paths, the blocked incident scenario, improved availability at dual-link sites and the new ability to detect and investigate internal traffic. He avoids putting a speculative monetary value on avoided incidents, instead noting the cost and disruption of the original guest malware incident as a reference point. Decision makers can then judge whether the outcomes justify the expense, and which elements, such as second links at further offices, are worth extending in future budgets.
Worked example
The outcome table is added to the dossier.
Synthetic case. Figures are synthetic.
Errors and limits
Short measurement periods can overstate availability.
Exercises
- EX136 — Reachability outcome. State the reachability outcome. Rubric: 3 = 30 to 7;; 2 = reduced;; 1 = vague;; 0 = none.
- EX137 — Availability limit. Why is three months a limitation? Rubric: 3 = short period may overstate;; 2 = short;; 1 = vague;; 0 = none.
- EX138 — Criteria met. Were all criteria met? Rubric: 3 = yes with evidence;; 2 = yes;; 1 = vague;; 0 = none.
Synthesis
The capstone meets its criteria within stated limits.
Continue studying — chapter 46 of 50
← Chapter 45 (draft) · Course chapter index · Chapter 47 (draft) →