Skip to main content

Workload and progression contract

Plan 512–743 focused hours. These are author estimates awaiting independent review and learner timing pilots. Record actual reading, implementation, debugging and assessment hours; extra prerequisite repair and optional depth are additional. Waiting for delayed recall is not study time.

Allocation without duplication​

  • Relational databases: 85–120 hours. Original modelling, SQL, normalization, migration and PostgreSQL plans/concurrency 63–88; executable relational core 22–32.
  • Storage engines: 90–130 hours. Original disk adapter/offset index, B+ trees, LSM policy, Bloom filters and plans 66–94; bounded-log core 24–36.
  • Replication/partitioning: 85–125 hours. Original topologies, quorum conflicts, skew, routing, failover and migration 63–93; replica-session core 22–32.
  • Transactions/consistency: 102–148 hours. Original isolation harness, WAL/checkpoints, consistency reasoning and sagas 66–94; outbox/inbox core 24–36; process/snapshot clinic 12–18.
  • Distributed fundamentals: 85–125 hours. Original fault models, detector, Raft and retry-safe API 65–95; logical-clock core 20–30.
  • Required privacy/retrieval strand: 35–50 hours. Inventory/lineage 6–8; deletion/retention 8–12; replay 9–13; retrieval benchmark 12–17.
  • Semester integration: 30–45 hours. Assemble/adapt project 12–18; checkpoint/cumulative review 6–9; exam 2.5 plus defense 2.5; repairs and delayed transfer 7–13.

Modules total 447–648 hours; strand and integration add 65–95. Extensions provide the implementation cores of their designated guided labs; they are not added again as duplicate labs. Original breadth and module quizzes remain in the module allocation. Reuse each artifact in several assessments, but record its creation time once. Integration budgets adaptation and synthesis, not five fresh implementations.

At 10 hours/week allow 52–75 weeks; at 15, 35–50; at 20, 26–38. Breaks and extensive repairs add time. Legacy weeks 60–69 are an ordering coordinate, not a ten-week completion promise.

Progression requirements​

All SQL1–SQL6, ST1–ST6, RP1–RP6, TX1–TX6, DB1–DB3 and DS1–DS6 require level 2 independently. All four privacy/retrieval categories D1–D4 also require level 2. Retain checkpoint 80%, exam 70% overall and 60% per major section, complete project and integrity requirements. Every condition applies; high averages do not compensate for missing evidence.

Use an evidence ledger: requirement, artifact, prediction, command/trace, observed outcome, limitation, score, repair and fresh retest. Score 0 missing/incorrect, 1 partial/prompted, 2 independently correct with evidence, 3 justified transfer. Any required score below 2 is repair required — incomplete. A repair plan alone does not permit progression.

Keep original real-system work visible: PostgreSQL query plans and isolation sessions; disk adapter and index rebuild; B+ tree/LSM/Bloom experiments; partition migration; WAL/checkpoint model; saga handling; detector/Raft/retry evidence. Passing the five supplied model labs is not completion of these broader requirements.

Sequence and incoming bridge​

Begin with schema and query semantics, then storage, replication, transactions and distributed reasoning. Review Semester 05 durable/acknowledged distinctions and Semester 04 ownership/partial I/O when those become weak. Build one service with an evolving evidence packet. Inventory its data early; add the strand's deletion/replay/retrieval path after the outbox and clock work. Use synthetic data so learning does not depend on personal records or external services.

Integrated defense and forward bridge​

During the 2.5-hour defense, trace one record through relational storage, outbox publication, replica visibility and a search projection. Delete it, restore an old snapshot and replay a stale event. Identify which mechanism prevents reappearance, which state still retains data, and what remains unknown. Then change the requirement from additive updates to ordered state replacement and explain which existing tests are insufficient.

Carry a decision packet into Semester 07: two measurable quality scenarios, an invariant, an observed failure, a recovery limit and a rejected alternative. Connect Semester 04 resource ownership and Semester 05 transport uncertainty to the service's data guarantees. After 48 hours, independently solve a new case in the weakest category. Preserve mistakes and corrections rather than rewriting the first attempt. Authored learner-ready material is not independently reviewed or battle-tested evidence.

Independent practice route and evidence reuse​

The following entry checks and changed-case practice make the existing module work inspectable. Use them within the current original-practice, lab and retention allowances; no full-book reading or new parallel project is added. An existing artifact can satisfy both an original exercise and this route when it demonstrates the same claim. Keep one evidence index linking each required gate to the exact artifact, actual run, assistance and reviewer status. Do not waive a topic simply because a different lab passes.

Keep authored material, author execution, learner attempts, independent review and delayed transfer as distinct statuses. Retain existing construction, security, HCI, AI-use, world-class and capstone strands and all per-gate minima. Real-system and human requirements remain pending until observed; planned observations are not completed evidence. Record actual study time before revising estimates.