Build a plan around the material you need to know

Database Systems courses differ in their SQL dialect, terminology and depth. Start with your course outline, lecture slides, prescribed reading and any available past questions. Use those sources to decide which topics are examinable; do not assume that a topic is required simply because it appears in a general database textbook.

This four-week plan assumes four 90-minute sessions each week: six hours in total. If you have less time, keep the session order and shorten the practice set. If your assessment is further away, repeat the cycle with new questions and more difficult examples. Each session should produce something you can check: a solved query, a dependency analysis, a transaction schedule or an explanation from memory.

Before the first session, sort your material into a small number of board sections. Keep definitions and diagrams beside the examples that make them usable. In MySummaries, a board for Database Systems might organise the material like this:

Database Systems Database Systems revisionStudy
Core concepts and problem solvingDatabase Systems revision5 sections · 3 columns
Relational model
  • A relation is a set of tuples; a candidate key uniquely identifies each tuple.
  • A foreign key references a candidate key in another relation.
  • Entity integrity: primary-key attributes are not NULL; referential integrity requires a matching referenced row or an allowed NULL.
Transactions4 due
  • ACID: atomicity, consistency, isolation and durability.
  • A lost update occurs when concurrent writes overwrite one another.
  • A conflict-serialisable schedule has an acyclic precedence graph.
SQL querying

Practise SELECT, JOIN, GROUP BY, HAVING, subqueries and NULL behaviour against small schemas. Check which SQL dialect your course uses.

Storage and query processing

Connect file organisation and index choices to the work a query performs. Review the B+ tree properties and cost assumptions used in your course.

Functional dependencies and normalisation
  • X → Y means each X value determines one Y value.
  • A relation is in 2NF if it is in 1NF and no non-prime attribute depends on part of a candidate key.
  • 3NF: for every non-trivial X → A, X is a superkey or A is prime.
  • BCNF: for every non-trivial X → Y, X is a superkey.

Week 1: establish what you know

Session 1 — 90 minutes: map the syllabus and diagnose. Spend 15 minutes listing the examinable headings from your course materials. For the remaining time, attempt a small sample of questions across the headings without looking at notes. Include a short SQL query, a key or dependency question, and a transaction or index question if those topics are on your syllabus. Mark each result as secure, uncertain or not yet understood. This is a starting diagnosis, not a score to optimise.

Session 2 — 90 minutes: relational model and ER modelling. Review entity sets, attributes, relationships, participation and cardinality using your course’s notation. Convert one ER diagram into relations. State primary keys and foreign keys, and explain how you handled a many-to-many relationship. Check whether weak entities or specialisation are in your course materials before spending time on them.

Session 3 — 90 minutes: SQL foundations. Work from a supplied schema. Write queries that select, filter, join and aggregate data. For every query, predict the result before running it if you have a database environment. Pay particular attention to the difference between filtering groups with HAVING and filtering rows with WHERE, and to how NULL affects comparisons. Verify dialect-specific syntax against course material.

Session 4 — 90 minutes: closed-book retrieval and correction. Recreate the key relational definitions and write one query from memory. Then check each answer against your notes and record the exact error: misunderstood relationship, wrong join condition, incorrect grouping or forgotten rule. Turn each meaningful error into a question for later retrieval.

Week 2: practise dependencies and normalisation

Session 1 — 90 minutes: functional dependencies. For each relation in a worked example, identify the stated dependencies, find candidate keys and justify your reasoning. Do not infer dependencies merely from a few sample rows; use the problem’s stated rules. Practise attribute closure if it is part of your course, showing each step so you can locate an error.

Session 2 — 90 minutes: normal forms. Use one relation and decompose it step by step. Identify the candidate keys and prime attributes first; then test the relevant normal-form definition. For each decomposition, check whether it is lossless and whether dependencies are preserved, using the methods taught in your course. Avoid memorising a decomposition without being able to explain why it is valid.

Session 3 — 90 minutes: mixed SQL and schema design. Design a small schema from a written scenario, then write queries against it. Check that your joins follow the keys you defined and that the schema avoids unintended duplication. Compare your result with your course examples rather than assuming a particular design is the only acceptable one.

Session 4 — 90 minutes: retrieve and repair. Attempt a fresh dependency problem and a fresh query without notes. Review wrong answers immediately. If the same distinction is still unclear, return to the relevant note, write a one-sentence rule in your own words and solve a new example. A due queue can help you decide which board needs retrieval practice first. In MySummaries, the queue could look like this:

Due today
Relational model4 due
SQL querying7 due
Normalisation5 due
TransactionsClear
Illustrative due counts show which Database Systems boards have cards waiting today.

Week 3: transactions, indexes and query processing

Session 1 — 90 minutes: transaction properties and anomalies. Define each ACID property, then work through short concurrent histories. Identify which operations conflict and what anomaly occurs. Be able to distinguish a dirty read, non-repeatable read and lost update using the definitions and examples in your course. Isolation-level behaviour can vary by DBMS, so learn the model specified by your materials rather than relying on a product-specific assumption.

Session 2 — 90 minutes: serialisability. For each schedule, draw the precedence graph using the conflict rules taught in class. Check all conflicting operation pairs, add directed edges between transactions, and test for a cycle. Explain what the result means; do not treat a graph with missing edges as proof of serialisability until you have checked every relevant pair.

Session 3 — 90 minutes: indexes and storage. Review the index structures actually covered in your course. For a B+ tree, practise tracing search, insertion or deletion if required, and state how internal nodes and leaves are used in the version taught. Then compare an indexed lookup with a scan using the cost model from your notes. The point is to connect the structure to the operation, not to claim that an index always makes a query faster.

Session 4 — 90 minutes: query evaluation and optimisation. Trace a query through the operators or execution plans used in your course. Practise explaining why selection pushdown or a different join order may reduce intermediate results, where applicable. Use the assumptions in the question: real optimisers and cost estimates depend on statistics, indexes and implementation.

Week 4: combine topics and target the gaps

Session 1 — 90 minutes: timed mixed set. Choose questions from different sections and give yourself a fixed 90-minute work period. Include both calculations or constructions and explanations. Do not assume the real assessment has a particular duration or format; use the timing only to practise working steadily. Mark against your course materials and note the reason for every lost point.

Session 2 — 90 minutes: highest-priority weak section. Rank errors by how often they recur and how much of the topic they affect. If dependency analysis is weak, solve several new examples and explain each step. If SQL is weak, vary the query requirement while keeping the schema fixed. In MySummaries, a picker can focus a short review on the section with the most evidence of difficulty:

New lecture
How long have you got?5 min10 min15 min20 min
TransactionsNormalisationSQL querying

Transactions · Struggling — getting 4 of 9 cards wrong and scoring 41% on the last review

Generate lecture ideas
Illustrative section heat directs a short listening review towards the weakest area.

Use a brief audio review as a prompt to recall the details, not as a replacement for solving problems. After listening, close the material and explain the key rule aloud or write a fresh example from memory.

Session 3 — 90 minutes: repair and re-test. Revisit the errors from Session 1, review the relevant notes, and attempt a different question of the same type. Check whether the correction holds when the numbers or schema change. Update your board only when you can state a useful rule or worked method clearly.

Session 4 — 90 minutes: final retrieval and next cycle. Reproduce core definitions, solve one representative problem from each major section and list any unresolved questions for your lecturer or course resources. Choose the next cycle’s priorities from your results, not from the topics you find most comfortable. A weak-area report can make that comparison visible. In MySummaries, a report for this plan might rank sections like this:

Where marks go missing
41%Transactions6×
63%Normalisation9×
88%Relational model7×
Illustrative results rank sections by recent performance and the number of times they have been practised.

How to keep the plan useful

Keep a short error log with three fields: the question type, what went wrong and what you will do differently next time. “Forgot normalisation” is too broad; “treated a non-prime attribute as prime when checking 3NF” tells you what to practise. Re-test the correction after a day or two, then again the following week.

When your course gives you past questions, use them to learn its terminology and level of detail, but do not memorise answers without understanding the method. If you are unsure whether a fact or topic is required, check the official course materials or ask the relevant teaching staff. Adjust the plan to the subject’s actual syllabus and your available time.

How MySummaries helps

MySummaries can organise your course notes into revision boards, turn sections into spaced-repetition cards and help you identify material that needs another pass. For this plan, use it to keep the SQL, normalisation and transaction topics connected to your own notes and practice results.