DBMS is one of the most predictable subjects in GATE CS. We counted every database question in all 23 official GATE CS papers from 2014 to 2026: DBMS averaged 7.1 marks per paper (7.1% of the 100-mark paper) and never fell below 5 or rose above 9. Among the core subjects, only Computer Networks (6 to 10) has as narrow a range. Within it, the balance has shifted. Normalization has roughly doubled since 2021, SQL questions have become less frequent, and transactions remain the single most-asked topic. This guide covers the official 2027 syllabus, topic-by-topic numbers, five worked questions from past papers and a study order built on that data.
How we counted: we classified all 1,495 GATE CS questions from 2014 to 2026 (23 papers, each totalling exactly 100 marks) by subject and topic, using the official papers as reproduced on Success Tracker's GATE CS PYQ archive. Hashing questions are counted under Algorithms, as the official syllabus lists them, not DBMS, because they test the hash table, not database indexing. Confirm the syllabus on the official GATE 2027 syllabus page (IIT Madras).
GATE CS 2027 DBMS Syllabus
Databases is Section 9 of the official GATE 2027 CS syllabus. The official text reads:
- ER-model
- Relational model: relational algebra, tuple calculus, SQL
- Integrity constraints, normal forms
- File organization, indexing (e.g., B and B+ trees)
- Transactions and concurrency control
This is word for word the same as the GATE 2026 CS syllabus. Two things are not listed: domain relational calculus and recovery algorithms such as ARIES. Recoverability of schedules is still asked, because it falls under transactions. If you are also taking GATE DA, note that its database syllabus is different. It includes data warehousing but not transactions (see our DBMS guide for GATE DA).
How Many Marks Does DBMS Carry in GATE CS?
The official question paper pattern fixes marks only for General Aptitude (15) and Engineering Mathematics (13). It does not give marks for individual CS subjects. This is what DBMS actually carried in each paper:
| Year | Marks per paper (set-wise) | % of the 100-mark paper | Questions |
|---|---|---|---|
| 2014 | 8 / 8 / 8 | 8% / 8% / 8% | 5 / 5 / 5 |
| 2015 | 6 / 6 / 6 | 6% / 6% / 6% | 4 / 4 / 4 |
| 2016 | 5 / 5 | 5% / 5% | 4 / 3 |
| 2017 | 8 / 8 | 8% / 8% | 5 / 5 |
| 2018 | 6 | 6% | 4 |
| 2019 | 8 | 8% | 5 |
| 2020 | 8 | 8% | 5 |
| 2021 | 8 / 8 | 8% / 8% | 5 / 5 |
| 2022 | 7 | 7% | 5 |
| 2023 | 5 | 5% | 3 |
| 2024 | 8 / 8 | 8% / 8% | 6 / 6 |
| 2025 | 8 / 9 | 8% / 9% | 5 / 5 |
| 2026 | 6 / 6 | 6% / 6% | 4 / 4 |
| Average (23 papers) | 7.1 | 7.1% (range 5–9%) | 4.6 |
What this means: sets in the same year almost always carry the same DBMS marks, and the range across 13 years is just 5 to 9. For about four to six weeks of study, DBMS reliably gives you around 7 marks, which makes it one of the best returns on time in the syllabus. For how DBMS compares with other subjects, see our GATE CSE subject-wise weightage analysis.
Topic-Wise Weightage: Where DBMS Marks Come From
We grouped the 106 DBMS questions by syllabus heading, with the average marks per paper across all 23 papers:
| Topic | Questions | Total marks | Avg marks per paper (23 papers) | Last asked |
|---|---|---|---|---|
| Transactions and concurrency control | 24 | 39 | 1.7 | 2026 |
| Relational model (algebra, tuple calculus, keys) | 25 | 38 | 1.7 | 2026 |
| Normalization (FDs, closures, normal forms) | 20 | 31 | 1.3 | 2026 |
| SQL | 16 | 27 | 1.2 | 2025 |
| File organization and indexing | 15 | 21 | 0.9 | 2026 |
| ER model | 6 | 7 | 0.3 | 2024 |
Looking at individual topics, the most-asked across all 23 papers were conflict serializability (12 questions, 22 marks), SQL aggregate and GROUP BY queries (9 questions, 16 marks), joins in relational algebra (8 questions, 15 marks), 1NF to 3NF (7 questions, 12 marks) and keys (7 questions, 8 marks).
Three patterns stand out:
- Normalization has doubled. It rose from about 0.9 to 1.9 marks per paper, and 2026 Set 1 had two normalization questions back to back. Functional dependencies, attribute closure and normal-form tests are now essential.
- SQL has roughly halved. It fell from about 1.5 to 0.8 marks per paper. When SQL does appear, it is usually a 2-mark "what does this query return" question with GROUP BY, HAVING or a nested subquery. Practise tracing queries on a small table by hand.
- Transactions are steady and heavily weighted. They score around 1.6 to 1.8 marks per paper, with conflict serializability alone making up 22 marks. Recent papers also test recoverability and ACID properties.
Question Types in DBMS
From 2021 to 2026, the 48 DBMS questions split into 21 MCQs, 16 MSQs and 11 numerical answer type (NAT) questions. They carried 32, 23 and 18 marks respectively, so about a third of DBMS marks now come from MSQs. A wrong 1-mark MCQ costs 1/3 mark and a wrong 2-mark MCQ costs 2/3 mark. MSQs and NATs have no negative marking, and an MSQ gives no partial credit (official marking rules).
DBMS MSQs are typically "which schedules are conflict serializable" or "which relations are in 3NF". Each option is an independent check, so work through them all one by one.
Topic by Topic: What GATE Asks and How to Prepare
1. Transactions and Concurrency Control
What gets asked: conflict serializability through precedence graphs, recoverable and cascadeless schedules, two-phase locking, ACID properties and the basics of log-based recovery.
How to prepare: two operations conflict only if they come from different transactions, touch the same data item and at least one is a write. Draw an edge for every conflicting pair, in time order, and check the graph for a cycle. For recoverability, ask whether any transaction commits after reading data written by a transaction that has not yet committed.
Recent PYQs: when two transactions conflict (2026), recoverability of a schedule (2025), ACID properties in a banking audit (2025, MSQ).
2. Relational Model: Algebra, Tuple Calculus and Keys
What gets asked: counting the tuples a relational algebra expression returns, joins (natural, theta, outer), division, which queries a tuple relational calculus expression represents, and counting superkeys or candidate keys.
How to prepare: for tuple-counting questions, build the intermediate tables on paper. Don't try to reason in your head. For TRC, translate the expression into plain English before you look at the options. For superkeys, use inclusion–exclusion, as in worked example 1.
Recent PYQs: tuple calculus over two relations (2026), teams and players TRC (2025).
3. Normalization: FDs, Closures and Normal Forms
What gets asked: attribute closure, candidate keys from a set of FDs, Armstrong's axioms, minimal cover, testing for 2NF, 3NF and BCNF, lossless join and dependency preservation.
How to prepare: always start by finding every candidate key. Attributes that never appear on the right-hand side of any FD must be in every key. Then test each FD against the definition. 3NF requires that for X → Y, either X is a superkey or Y is a prime attribute. BCNF requires X to be a superkey.
Recent PYQs: functional dependencies (2026, MSQ), normalization statements (2026, MSQ), which axioms prove the union rule (2026).
4. SQL
What gets asked: the output of queries with GROUP BY and HAVING, aggregate functions, nested and correlated subqueries, joins in SQL, and how NULL values behave.
How to prepare: learn the logical evaluation order: FROM → WHERE → GROUP BY → HAVING → SELECT → ORDER BY. Most SQL traps come from treating HAVING like WHERE, or from forgetting that aggregates ignore NULLs (except COUNT(*)).
Recent PYQs: students query (2025, MSQ), sports-league subquery (2025, NAT), employees GROUP BY (2021).
5. File Organization and Indexing
What gets asked: B and B+ tree insertion and splitting, the minimum and maximum number of keys or nodes, tree order from block size, dense versus sparse indexes, and primary versus secondary indexes.
How to prepare: practise B+ tree insertions until splitting becomes routine, and be precise about the definition of order, which differs between books. Read the question's own definition every time. For order-from-block-size questions, write out the inequality: n × pointer size + (n − 1) × key size ≤ block size.
Recent PYQs: B+ tree insertion (2025, MSQ), B+ tree search path (2025, NAT).
6. ER Model
What gets asked: weak entity sets and identifying relationships, cardinality and participation, and turning ER diagrams into tables. Most of these are 1-mark questions.
How to prepare: this is the smallest topic (7 marks in 13 years). Cover it with the lecture and the PYQs, and spend the time you save on normalization.
Recent PYQs: weak entities (2024), ER specification (2024).
Worked Examples from Official GATE CS Papers
Each answer below matches the official answer key, as reproduced on the linked question page.
Example 1: Counting superkeys (GATE CS 2026 Set 1, Q65, 2 marks, NAT)
Question: R(A, B, C, D) has exactly two candidate keys, {A, B} and {A, C}. How many superkeys does R have?
Solution: a superkey is any set that contains a candidate key. Sets containing {A, B} can add any subset of {C, D}, which gives 2² = 4. Sets containing {A, C} can add any subset of {B, D}, which also gives 4. Sets containing both, meaning sets containing {A, B, C}, can add D or not, which gives 2. By inclusion–exclusion: 4 + 4 − 2 = 6. Practise this question.
Example 2: Join then select (GATE CS 2024 Set 1, Q35, 1 mark, NAT)
Question: R(A, B) = {(10,20), (20,30), (30,40), (30,50), (50,95)} and S(A, C) = {(10,90), (30,45), (40,80)}. How many tuples does σB<C(R ⋈R.A=S.A S) return?
Solution: the join matches A = 10, giving (10, 20, 90), and A = 30, giving (30, 40, 45) and (30, 50, 45). A = 20 and A = 50 have no partner in S. Now apply B < C: 20 < 90 holds, 40 < 45 holds and 50 < 45 fails. The answer is 2. Practise this question.
Example 3: Conflict serializability (GATE CS 2025 Set 2, Q53, 2 marks, MSQ)
Question: T1 = R1(X), W1(Y), R1(X), W1(X) and T2 = W2(X), W2(Y). Which interleavings are conflict serializable?
Solution: a schedule is conflict serializable only if every conflict points the same way. Option B is W2(X), R1(X), W2(Y), W1(Y), R1(X), COMMIT(T2), W1(X), COMMIT(T1). Here T2 writes X before T1 reads it, T2 writes Y before T1 writes it, and T2 writes X before T1 writes it. Every edge is T2 → T1, so there is no cycle and B is serializable. In options A, C and D, one conflict goes T1 → T2 (T1 reads X or writes Y before T2 touches it) while another goes T2 → T1, which forms a cycle. The answer is B only. Practise this question.
Example 4: Testing 3NF (GATE CS 2025 Set 2, Q46, 2 marks, MSQ)
Question: R1(A, B, C, D, E) has D → E, EA → B and EB → C. R2(A, B, C, D) has A → D, A → B and C → A. Which relations are in 3NF?
Solution: in R1, A and D never appear on the right-hand side, and (AD)+ = {A, B, C, D, E}, so AD is the only key. D → E has a left side that is not a superkey and a right side that is not prime, so R1 is not in 3NF. In R2, C is on no right-hand side, and C+ = {A, B, C, D}, so C is the only key. A → D fails the same test, so R2 is not in 3NF either. The correct options are "R1 is NOT in 3NF" and "R2 is NOT in 3NF". Practise this question.
Example 5: Dense versus sparse indexes (GATE CS 2026 Set 2, Q46, 2 marks)
Question: S1 says a hash index must be dense. S2 says a B+ tree index can be sparse. Which is true?
Solution: a hash index cannot be sparse. Hashing does not preserve order, so a key that isn't in the index cannot be found by searching for its nearest neighbour. A B+ tree used as a clustering index on a sorted file can store one entry per block, which makes it sparse. Both S1 and S2 are true. Practise this question.
Study Order and Time Plan
DBMS takes roughly 4 to 5 weeks at 2 hours a day. This order builds each topic on the one before it:
- ER model and keys (3 days): the vocabulary for everything else.
- Relational algebra and tuple calculus (5–6 days): SQL is much easier once algebra is clear.
- SQL (4 days): focus on GROUP BY, HAVING, nested queries and NULLs.
- Functional dependencies and normalization (6–7 days): the fastest-growing topic, so give it the most practice.
- Transactions and concurrency control (5–6 days): precedence graphs, recoverability and 2PL.
- Indexing and B+ trees (4 days): insertions, splitting and order calculations.
- PYQ sweep (3 days): all 106 DBMS questions from 2014 to 2026 on the GATE CS PYQ archive.
For where DBMS fits among all the subjects, see the GATE CSE 2027 month-by-month study plan.
Best Books for GATE CS DBMS
| Book | Use it for |
|---|---|
| Silberschatz, Korth & Sudarshan, Database System Concepts (7th ed., db-book.com) | The primary text for the whole syllabus. |
| Elmasri & Navathe, Fundamentals of Database Systems | Extra practice on normalization and ER modelling. |
| Ramakrishnan & Gehrke, Database Management Systems | Indexing and concurrency control. |
Skip the chapters on distributed and parallel databases, query processing internals and NoSQL, which are not on the GATE CS syllabus. For other subjects, see our best books for GATE CSE 2027.
Free Video Lectures: DBMS for GATE CS
Our DBMS for GATE CS and DA playlist on YouTube is taught by Sriniwas Paliwal (AIR 90, GATE CS 2019). It follows the same order as the study plan above:
- ER model and keys: entities and attributes, relationships and constraints, weak entities, super, candidate and primary keys, foreign keys and integrity constraints
- Relational algebra and TRC: fundamentals, joins, division, tuple relational calculus, RA and TRC PYQs
- Functional dependencies and normalization: FDs and attribute closure, Armstrong's axioms, minimal cover, 1NF and 2NF, with 3NF and BCNF coming next
The full course, covering all subjects with mock tests, is on the GATE CS course page.
Common Mistakes in GATE DBMS
- Counting read–read pairs as conflicts when drawing precedence graphs. A conflict needs at least one write.
- Testing 3NF before finding every candidate key. Whether an attribute is prime depends on knowing all the keys.
- Treating HAVING like WHERE. HAVING filters groups after aggregation.
- Forgetting that a natural join drops rows with no match in tuple-counting questions (Example 2).
- Mixing up B+ tree order definitions between books. Use the definition the question gives.
Key Takeaways
- DBMS averaged 7.1 marks per paper (7.1% of the paper) across 23 official GATE CS papers (2014–2026), always between 5 and 9.
- Transactions (39 marks) and the relational model (38 marks) have the most marks overall. Conflict serializability is the single most-asked topic.
- Normalization roughly doubled to about 1.9 marks per paper since 2021, while SQL fell to about 0.8.
- About a third of DBMS marks since 2021 came from MSQs. Check every option.
Frequently Asked Questions
How many marks does DBMS carry in GATE CS?
The official pattern does not fix marks per CS subject. Across all 23 official GATE CS papers from 2014 to 2026, DBMS averaged 7.1 marks per paper, which is 7.1% of the 100-mark paper, and always stayed between 5 and 9.
Which DBMS topics are most important for GATE CS 2027?
Transactions (especially conflict serializability), the relational model and normalization. Normalization has grown the most, from about 0.9 to 1.9 marks per paper since 2021.
Is DBMS easy to score in GATE CS?
Yes, it is one of the most reliable scoring subjects. The marks are steady and the question types repeat. Of the 106 DBMS questions, Success Tracker rates 32 as easy, 69 as medium and only 5 as hard.
Which book is best for DBMS for GATE CS?
Database System Concepts by Silberschatz, Korth and Sudarshan covers the full syllabus. Add Elmasri and Navathe for extra normalization practice.
Is the DBMS syllabus the same for GATE CS and GATE DA?
No. GATE CS includes transactions and concurrency control, while GATE DA includes data warehousing instead. Our GATE DA DBMS guide covers the DA version, and our GATE DA vs GATE CS comparison explains the overlap.
Where can I practise GATE CS DBMS PYQs?
Every GATE CS question from 2014 to 2026 is free on Success Tracker's GATE CS PYQ archive, with the official answer and a worked solution.
Next in This Series
- Discrete Mathematics for GATE CS 2027: weightage, PYQs and plan
- GATE CSE subject-wise weightage
- GATE CSE 2027 important topics
- GATE CSE previous year question papers
- GATE CS 2027 complete guide
Sources
- GATE 2027 CS Syllabus (PDF), IIT Madras: Section 9, Databases
- GATE 2026 CS Syllabus (PDF), IIT Guwahati
- GATE 2027 Question Paper Pattern, IIT Madras: marks distribution and negative marking
- Database System Concepts, 7th edition: official companion site
- GATE 2027 Information Brochure, IIT Madras: Section 7.3, Table 8 (Engineering Mathematics 13 marks in CS)
- Official GATE CS question papers and answer keys, IIT Madras download page
- Official GATE CS question papers and answer keys, 2014–2026 (23 papers), as reproduced on Success Tracker. Weightage figures are The ML Hub's own count, done on 25 October 2026.